emi noise reduction techniques for high frequency … · 2018-05-23 · forms a wheatstone bridge...
TRANSCRIPT
EMI Noise Reduction Techniques
for High Frequency Power Converters
Yuchen Yang
Dissertation submitted to the Faculty of the
Virginia Polytechnic Institute and State University
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in
Electrical Engineering
APPROED
Fred C. Lee (Chair)
Qiang Li
Jaime De La Ree
Virgilio Centeno
Steve C. Southward
April 16th, 2018
Blacksburg, Virginia
Keywords: Wide-band-gap devices, high frequency, EMI, Magnetic
integration
© 2018, Yuchen Yang
EMI Noise Reduction Techniques
for High Frequency Power Converters
Yuchen Yang
Abstract
Switch mode power supplies are widely used in different applications. High efficiency and
high power density are two driving forces for power supply systems. However, high dv/dt and di/dt
in switch mode power supplies will cause severe EMI noise issue. In a typical front-end converter,
the EMI filter usually occupies 1/3 to 1/4 volume of total converter. Hence, reducing the EMI
noise of power converter can help reduce the volume of EMI filter and improving the total power
density of the converter.
The EMI noise can be separated as differential mode (DM) noise and common mode (CM)
noise. For off-line switch mode power supplies, DM noise is dominated by PFC converter. CM
noise is a more complicated issue. It is contributed by both PFC converter and DC/DC converter.
The DM noise is contributed by input current ripple. Therefore, one method to reduce DM noise
is interleaving. There are three methods to reduce CM noise: symmetry, balance and shielding.
The idea of symmetry concept is generating another dv/dt source to cancel the original dv/dt source.
However, this method is very difficult to achieve and usually has more loss. The balance technique
forms a Wheatstone bridge circuit to minimize the CM noise. However, the balance technique
cannot achieve very good attenuation at high frequency due to parasitics. Shielding technique is
very popular in isolated DC-DC converters to reduce CM noise. However, the previous shielding
method requires precise control of parasitic capacitance and dv/dt. It is very difficult to achieve
good CM noise attenuation in mass production.
In this dissertation, a novel one-layer shielding method for PCB winding transformer is
provided. This shielding technique can block CM noise from primary side and also cancel the CM
noise from secondary side. In addition, shielding does not increase the loss of converter too much.
Furthermore, this shielding technique can be applied to matrix transformer structure. For matrix
transformer LLC converter, the inter-winding capacitor is very large and will cause severe CM
noise problem. By adding shielding layer, CM noise has been greatly reduced. Although flyback
and LLC resonant converter are used as examples to demonstrate the concept, the novel shielding
technique can also be applied to other topologies that have similar transformer structure.
With Wide-band-gap power devices, the switching frequency of power converter can be
pushed 10 times higher than traditional Si based converters. This provides an opportunity to use
PCB winding magnetics. In order to reduce the switching loss, critical conduction mode is used in
PFC converter. Because of high AC current in the inductor winding, litz wire was used to build
the inductor. However, with coupled inductor concept and the proposed winding structure, CRM
inductor is integrate into PCB winding for the first time. Furthermore, balance technique is applied
to reduce CM noise for PFC converter. With PCB winding, the balance technique has better high
frequency performance. The PCB winding inductor can achieve high power density, high
efficiency and automated manufacture.
Traditionally, two-stage EMI filter was utilized to achieve required EMI noise attenuation.
With the developed high frequency, low EMI noise converter, single-stage EMI filter can be
applied. However, there are self-parasitic and mutual parasitic components to impact the filter
performance on high frequency. The near-field measurement is utilized to visualize the magnetic
flux near those filter components. Thus, a better filter design and layout can be achieved to have
better high frequency performance.
EMI Noise Reduction Techniques
for High Frequency Power Converters
Yuchen Yang
General Audience Abstract
Switch mode power supplies are widely used in different applications. High efficiency and
high power density are two driving forces for power supply systems. In a world full of electronic
devices, it is very important that these devices can work properly in a complicated electromagnetic
environment. Thus, electromagnetic compatibility (EMC) is a significant characteristic of
electronic devices. However, high dv/dt and di/dt in switch mode power supplies will cause severe
EMI noise issue. In a typical front-end converter, the EMI filter usually occupies 1/3 to 1/4 volume
of total converter. Hence, reducing the EMI noise of power converter can help reduce the volume
of EMI filter and improving the total power density of the converter. In this dissertation, several
methods to reduce EMI noise are proposed and analyzed. First, the shielding method for PCB
winding transformer is proposed. It can effectively reduce EMI noise at wide frequency range.
Second, balance technique is applied to reduce EMI noise of PFC converter. Traditionally, two-
stage EMI filter was utilized to achieve required EMI noise attenuation. With the developed high
frequency, low EMI noise converter, single-stage EMI filter can be applied. However, there are
self-parasitic and mutual parasitic components to impact the filter performance on high frequency.
The near-field measurement is utilized to visualize the magnetic flux near those filter components.
Thus, a better filter design and layout can be achieved to have better high frequency performance.
v
To My Family
My parents: Shijun Yang and Ying Lu
My parents-in-law: Jiquan Li and Qiusha Dai
My wife: Zhuxuan Li
vi
Acknowledgments
First and foremost I would like to thank my advisor, Dr. Fred C. Lee for his guidance,
encouragement and education. It is him who opens the door of power electronics for me and show
me the right way to do research. He is not only a wise and farsighted advisor in academic, but also
a mentor in life. He pointed out my character weaknesses and helped me become a better person.
His profound knowledge, rich experience, rigorous attitude, and challenging spirit have been a
great value for me. It is most honored to be advised by Dr. Lee. The experience of studying at
CPES is a life-changing experience.
I am very grateful to my other committee members Dr. Qiang Li, Dr. Jaime De La Ree, Dr.
Virgilio Centeno, and Dr. Steve Southward. I still remember the weekly meeting with Dr. Li when
I first came to CPES. He taught me how to start a research and how to express my ideas. Thanks
for all the constructive and helpful suggestions offered by Dr. Li.
My very special thanks go to my colleagues Dr. Zhengyang Liu and Mr. Bin Li. As a team,
we have worked on several projects. I cannot accomplish so much without their help.
I really appreciate the discussions and suggestions from Dr. Dushan Boroyevich, Dr. Khai
Ngo, Dr. Rolando Burgos, and Dr. Guo-Quan Lu in High Density Integration (HDI) group.
I really want to thank all my colleagues in the Power Management Consortium (PMC) group:
Dr. David Reusch, Dr. Mingkai Mu, Dr. Yingyi Yan, Dr. Weiyi Feng, Dr. Daocheng Huang, Dr.
Zijian Wang, Mr. Shu Ji, Mr. Haoran Wu, Dr. Yipeng Su, Mr. Wei Zhang, Mr. Li Jiang, Dr. Shuilin
Tian, Dr. Pei-Hsin Liu, Dr. Dongbin Hou, Dr. Yang Jiao, Mr. Sizhao Lu, Mr. Chao Fei, Mr.
Xuebing Chen, Ms. Yincan Mao, Mr. Zhongsheng Cao, Mr. Syed Bari, Mr. Tao Liu, Ms. Virginia
vii
Li, Mr. Junjie Feng, Mr. Yi-Hsun Hsieh, Mr. Mohamed Ahmed, Mr. Yadong Lyu, Mr. Rimon
Gadelrab, Mr. Shishuo Zhao, Mr. Shuo Wang, Mr. Yinsong Cai, Mr. Feiyang Zhu. Mr. Owen Jong,
Mr. Ahmed Nabih, Dr. Xinke Wu, Dr. Xin Ming, Dr. Weijing Du, Dr. Shuojie She, Mr. Yan-Cun
Li, Dr. Kenichiro Tanaka, Dr. Minfan Fu, Mr. Furong Xiao, Mr. Guo Xu, and Dr. Rong Xu. My
thanks also go to many of the other students and visiting scholars I have met in CPES.
I sincerely thank all the wonderful CPES staffs: Ms. Teresa Shaw, Mr. Douglas Sterk, Mr.
David Gilham, Ms. Linda Gallagher, Ms. Teresa Rose, Ms. Marianne Hawthorne, Ms. Linda Long,
Mr. Igor Cvetkovic, and Ms. Lauren Shutt for their help and support.
In addition, this work was supported by CPES Power Management Consortium, and the
Engineering Research Center Shared Facilities supported by the National Science Foundation
under NSF Award Number EEC-9731677. Any opinions, findings and conclusions or
recommendations expressed in this material are those of the author and do not necessarily reflect
those of the National Science Foundation.
The information, data, or work presented herein was funded in part by the Office of Energy
Efficiency and Renewable Energy (EERE), U.S. Department of Energy, under Award Number
DE-EE0006521 with North Carolina State University, PowerAmerica Institute.
The information, data, or work presented herein was funded in part by an agency of the United
States Government. Neither the United States Government nor any agency thereof, nor any of
their employees, makes any warranty, express or implied, or assumes any legal liability or
responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product,
or process disclosed, or represents that its use would not infringe privately owned rights.
Reference herein to any specific commercial product, process, or service by trade name, trademark,
manufacturer, or otherwise does not necessarily constitute or imply its endorsement,
viii
recommendation, or favoring by the United States Government or any agency thereof. The views
and opinions of authors expressed herein do not necessarily state or reflect those of the United
States Government or any agency thereof.
This material is partially based upon work supported by the U.S. Department of Energy, Office
of Energy Efficiency and Renewable Energy, under the project of High Efficiency High Density
GaN Based 6.6kW Bi-Directional On-board Charger for PEVs, through Delta Products Corp,
under contract number DE-EE0006834.
ix
Table of Contents
Chapter 1. Introduction ................................................................................................................. 1
1.1 Introduction to EMI and EMC .............................................................................................. 1
1.2 EMI Measurement................................................................................................................. 3
1.3 Conducted EMI Noise in Switch Mode Power Supplies ...................................................... 5
1.4 EMI Noise Reduction Methods for Switch Mode Power Supplies ....................................... 8
1.5 Proposed Dissertation Outline ............................................................................................ 21
Chapter 2. Shielding Technique for Isolated DC/DC Converter ............................................. 23
2.1 PCB Winding Transformer with Shielding Technique ....................................................... 23
2.2 Matrix Transformer with Shielding Technique ................................................................... 31
2.3 Conclusions ......................................................................................................................... 40
Chapter 3. Balance Technique for Coupled Inductors Integrated into PCB.......................... 42
3.1 PCB Winding Negative Coupled Inductor with Balance Technique .................................. 43
3.1.1 1.2kW GaN Based Totem-pole Bridgeless PFC above MHz ............................... 43
3.1.2 Negative Coupled Inductor in MHz CRM PFC Converter ................................... 49
3.1.3 PCB Winding Inductor Design ............................................................................. 54
3.1.4 Balance Technique for Coupled Inductor ............................................................. 60
3.1.5 Hardware Demonstration ...................................................................................... 68
3.2 PCB Winding Positive Coupled Inductor with Balance Technique ................................... 72
3.2.1 6.6kW SiC Based PFC Converter for On-board Battery Charger ........................ 72
3.2.2 Negative Coupled Inductor for 6.6kW PFC .......................................................... 77
x
3.2.3 Positive Coupled Inductor for 6.6kW PFC ........................................................... 80
3.2.4 PCB Winding Inductor Design ............................................................................. 84
3.2.5 Improved PCB Winding Inductor Design ............................................................. 86
3.2.6 Balance Technique to Reduce CM Noise ............................................................. 93
3.3 Conclusions ......................................................................................................................... 96
Chapter 4. EMI Filter Design Considerations ........................................................................... 98
4.1 Single Stage EMI Filter ..................................................................................................... 102
4.2 Parasitic of DM Filter ....................................................................................................... 104
4.3 Near-field Analysis for Mutual Parasitic Cancellation ..................................................... 107
4.4 Reduction of ESL of Capacitor Branch ............................................................................ 118
4.5 Reduction of Mutual Coupling between Two Capacitors ................................................. 122
4.6 Single Stage EMI Filter Demonstration in 1 kW Server Power Supply ........................... 126
4.7 Conclusions ....................................................................................................................... 128
Chapter 5. Conclusions and Future Work ............................................................................... 129
Reference 132
xi
List of Figures
Fig. 1.1 EMI standard EN55022 Class B for conducted noise ..................................................................... 2
Fig. 1.2 FCC measurement setup .................................................................................................................. 3
Fig. 1.3 Internal circuit of a LISN ................................................................................................................. 4
Fig. 1.4 Practical noise measurement circuit ................................................................................................ 4
Fig. 1.5 Structure of low power converter .................................................................................................... 5
Fig. 1.6 Front-end converter for server and telecom ..................................................................................... 6
Fig. 1.7 State-of-the-art front end converter ................................................................................................. 6
Fig. 1.8 DM noise propagation path ............................................................................................................. 7
Fig. 1.9 CM noise propagation path .............................................................................................................. 7
Fig. 1.10 Full bridge converter...................................................................................................................... 9
Fig. 1.11 Waveforms of V1 and V2 ............................................................................................................... 9
Fig. 1.12 Boost converter with symmetry technique .................................................................................. 10
Fig. 1.13 Balanced boost converter ............................................................................................................. 11
Fig. 1.14 Replace the MOSFET with a voltage source ............................................................................... 12
Fig. 1.15 CM noise model of balanced boost PFC converter ..................................................................... 12
Fig. 1.16 Wheatstone bridge structure ........................................................................................................ 13
Fig. 1.17 CM noise reduction of boost converter with balance condition achieved at low frequency [8] .. 14
Fig. 1.18 Balance technique with coupled inductors [8] ............................................................................. 15
Fig. 1.19 CM noise of balanced boost converter with inductor coupling [8] ............................................. 15
Fig. 1.20 Flyback converter and CM noise path ......................................................................................... 16
Fig. 1.21 Series Shielding method .............................................................................................................. 17
Fig. 1.22 Voltage distribution of P3 and S .................................................................................................. 18
Fig. 1.23 Misalignment between primary and secondary windings ............................................................ 18
Fig. 1.24 Partial shielding method .............................................................................................................. 19
xii
Fig. 1.25 CM noise model of partial shielding ........................................................................................... 20
Fig. 1.26 Waveform of VA and VD .............................................................................................................. 20
Fig. 2.1 Conventional flyback converter transformer ................................................................................. 24
Fig. 2.2 Size comparison between 1 MHz PCB winding transformer and 100 kHz litz wire transformer . 25
Fig. 2.3 Cross-section view of PCB winding transformer .......................................................................... 25
Fig. 2.4 CM noise measurement for PCB winding flyback converter ........................................................ 26
Fig. 2.5 Flyback converter with shielding layer .......................................................................................... 27
Fig. 2.6 Detailed PCB layout with jumper connection ............................................................................... 27
Fig. 2.7 Shielding has a 180º rotation with secondary winding .................................................................. 29
Fig. 2.8 3-D structure of shielding and secondary winding ........................................................................ 29
Fig. 2.9 Voltage distribution of shielding and secondary winding ............................................................. 29
Fig. 2.10 Shielding serving as part of primary winding .............................................................................. 30
Fig. 2.11 Transformer cross-section view with shielding ........................................................................... 30
Fig. 2.12 CM noise reduction for flyback converter with shielding ........................................................... 31
Fig. 2.13 Conventional LLC converter ....................................................................................................... 33
Fig. 2.14 LLC resonant with matrix transformer ........................................................................................ 34
Fig. 2.15 Matrix transformer structure ........................................................................................................ 35
Fig. 2.16 Mount secondary devices and output capacitors on top of secondary winding ........................... 35
Fig. 2.17 CM noise measurement result for LLC resonant converter with matrix transformer .................. 36
Fig. 2.18 Adding shielding into matrix transformer ................................................................................... 37
Fig. 2.19 Transformer structure with shielding ........................................................................................... 37
Fig. 2.20 3D structure of secondary winding and shielding ....................................................................... 38
Fig. 2.21 Voltage distribution of shielding and secondary winding ........................................................... 39
Fig. 2.22 CM noise measurement result for matrix transformer with shielding ......................................... 39
Fig. 2.23 Efficiency comparison between shielding version and original version ..................................... 40
Fig. 3.1 A typical AC-DC converter ........................................................................................................... 42
xiii
Fig. 3.2 Bridgeless PFC converter topologies ............................................................................................. 44
Fig. 3.3 Measured switching loss distribution of different GaN device samples ........................................ 45
Fig. 3.4 Critical mode operation ................................................................................................................. 45
Fig. 3.5 Two-phase interleaved totem-pole PFC converter ........................................................................ 46
Fig. 3.6 Waveforms of two-phase interleaved totem-pole PFC converter .................................................. 46
Fig. 3.7 DM noise measurement result ....................................................................................................... 47
Fig. 3.8 Prototype of two-phase interleaved 1.2 kW totem-pole PFC converter ........................................ 47
Fig. 3.9 Switching frequency variation of CRM PFC converter ................................................................. 48
Fig. 3.10 Non-ZVS at high input voltage .................................................................................................... 49
Fig. 3.11 Interleaved totem-pole PFC converter with coupled inductor ..................................................... 50
Fig. 3.12 Negative coupled inductor structure demonstration .................................................................... 50
Fig. 3.13 Inductor current waveforms of coupled inductor ........................................................................ 51
Fig. 3.14 Steady-state inductance variance during half-line cycle .............................................................. 52
Fig. 3.15 Switching frequency variance of coupled inductor during half -line cycle ................................. 52
Fig. 3.16 Coupling can help achieve ZVS .................................................................................................. 54
Fig. 3.17 Coupling can help extend ZVS range .......................................................................................... 54
Fig. 3.18 Size comparison between 100 kHz inductor and 1 MHz inductor .............................................. 55
Fig. 3.19 Core loss density of 3F45 and P61 .............................................................................................. 55
Fig. 3.20 Coupled inductor in UI core ........................................................................................................ 57
Fig. 3.21 Coupled inductor with interleaved winding................................................................................. 58
Fig. 3.22 Simulation of flux distribution with large fringing flux .............................................................. 58
Fig. 3.23 Coupled inductor structure III ..................................................................................................... 59
Fig. 3.24 Simulation result of avoiding fringing flux ................................................................................. 59
Fig. 3.25 CM noise comparison between GaN based PFC and Si based PFC ............................................ 60
Fig. 3.26 Balancing technique for interleaved totem-pole PFC converter with coupled inductor .............. 61
Fig. 3.27 CM noise model of PFC converter with balance ......................................................................... 62
xiv
Fig. 3.28 CM noise model of one noise source ........................................................................................... 62
Fig. 3.29 Circuit structure of improved balance technique ......................................................................... 63
Fig. 3.30 Inductor structure demonstration of improved balance technique ............................................... 64
Fig. 3.31 CM noise model of improved balance technique ........................................................................ 64
Fig. 3.32 Effect of only noise source VN1 ................................................................................................... 64
Fig. 3.33 Magnetic circuit of coupled inductor with balance ..................................................................... 66
Fig. 3.34 Equivalent circuit of coupled inductor with balance ................................................................... 66
Fig. 3.35 Simplified equivalent circuit of coupled inductor with balance .................................................. 66
Fig. 3.36 Current waveforms verification ................................................................................................... 67
Fig. 3.37 Cross-section view of coupled inductor with balance technique ................................................. 67
Fig. 3.38 Add capacitor to maintain balance condition. ............................................................................. 68
Fig. 3.39 Winding arrangement for proposed PCB winding inductor ........................................................ 69
Fig. 3.40 Comparison between discrete inductor and integrated inductor .................................................. 70
Fig. 3.41 Hardware photo of 1.2kW CRM PFC with PCB winding inductor ............................................ 71
Fig. 3.42 Measured efficiency of 1.2kW PFC converter ............................................................................ 71
Fig. 3.43 CM noise measurement result ..................................................................................................... 72
Fig. 3.44 Target of high efficiency high density bidirectional OBC .......................................................... 73
Fig. 3.45 Battery charging profile ............................................................................................................... 74
Fig. 3.46 Circuit structure for 6.6kW on-board battery charger ................................................................. 75
Fig. 3.47 Gain range of proposed CLLC converter .................................................................................... 75
Fig. 3.48 Switching frequency variation during half line cycle .................................................................. 76
Fig. 3.49 Negative coupled inductor for 6.6kW PFC ................................................................................. 77
Fig. 3.50 Steady state inductance for 6.6kW PFC with negative coupled inductor .................................... 78
Fig. 3.51 Switching frequency for 6.6kW PFC with negative coupled inductor ........................................ 79
Fig. 3.52 Duty cycle range for 1 kW server power supple ......................................................................... 80
Fig. 3.53 Duty cycle range for 6.6 kW OBC .............................................................................................. 80
xv
Fig. 3.54 Positive coupled inductor for 6.6kW PFC ................................................................................... 81
Fig. 3.55 Positive coupled inductor structure demonstration ...................................................................... 82
Fig. 3.56 Typical waveforms of positive coupled inductor ........................................................................ 82
Fig. 3.57 Steady state inductance for 6.6kW PFC with positive coupled inductor ..................................... 83
Fig. 3.58 Switching frequency for 6.6kW PFC with positive coupled inductor ......................................... 83
Fig. 3.59 Litz wire inductors for OBC ........................................................................................................ 84
Fig. 3.60 Conventional PCB winding inductor ........................................................................................... 85
Fig. 3.61 Proposed PCB winding inductor with interleave ......................................................................... 86
Fig. 3.62 Prototype of 6.6kW on-board battery charger ............................................................................. 87
Fig. 3.63 Measured PFC converter efficiency ............................................................................................ 87
Fig. 3.64 Experimental waveforms of 6.6kW PFC converter ..................................................................... 88
Fig. 3.65 Path of current spikes and ringing ............................................................................................... 89
Fig. 3.66 Inductor optimization parameters ................................................................................................ 89
Fig. 3.67 Inductor loss breakdown for inductor design .............................................................................. 90
Fig. 3.68 Contour of total loss and footprint ............................................................................................... 91
Fig. 3.69 Prototype Gen. 2 of 6.6kW on-board battery charger .................................................................. 91
Fig. 3.70 Experimental waveforms of Gen. 2 6.6kW PFC converter ......................................................... 92
Fig. 3.71 Improved efficiency of Gen. 2 hardware ..................................................................................... 93
Fig. 3.72 Balance technique for positive coupled inductor ......................................................................... 94
Fig. 3.73 CM noise model of balanced PFC converter ............................................................................... 94
Fig. 3.74 Effect of noise source VS1 ............................................................................................................ 94
Fig. 3.75 PCB winding positive coupled inductor with balance ................................................................. 96
Fig. 3.76 CM noise reduction by balance technique ................................................................................... 96
Fig. 4.1 State-of-the-art server power supply.............................................................................................. 98
Fig. 4.2 Two-stage EMI filter structure ...................................................................................................... 99
Fig. 4.3 EMI noise measurement result for 1 kW server power supply .................................................... 101
xvi
Fig. 4.4 One-stage EMI filter structure ..................................................................................................... 102
Fig. 4.5 EMI noise measurement results ................................................................................................... 103
Fig. 4.6 Equivalent circuit for DM filter ................................................................................................... 104
Fig. 4.7 Equivalent circuit for CM filter ................................................................................................... 104
Fig. 4.8 Insertion voltage gain for ideal DM filter .................................................................................... 105
Fig. 4.9 DM filter with self parasitics ....................................................................................................... 105
Fig. 4.10 Insertion voltage gain for actual DM filter ................................................................................ 106
Fig. 4.11 DM filter with self and mutual parasitics .................................................................................. 106
Fig. 4.12 Mutual coupling between inductor and capacitor ...................................................................... 107
Fig. 4.13 Near-field flux measurement setup ............................................................................................ 108
Fig. 4.14 DM inductor structure and CM inductor structure .................................................................... 108
Fig. 4.15 Near-field flux measured with 1MHz excitation ....................................................................... 109
Fig. 4.16 Flux flow direction of 1 MHz excitation ................................................................................... 110
Fig. 4.17 Near field measurement result of 10 MHz................................................................................. 111
Fig. 4.18 Flux flow direction of 10 MHz excitation ................................................................................. 111
Fig. 4.19 Near field measurement result of 15 MHz................................................................................. 112
Fig. 4.20 Flux flow direction of 15 MHz excitation ................................................................................. 112
Fig. 4.21 Near field measurement result of 25 MHz................................................................................. 113
Fig. 4.22 Flux flow direction of 25 MHz excitation ................................................................................. 114
Fig. 4.23 Different capacitor position with horizontal inductor ................................................................ 115
Fig. 4.24 Traditional EMI filter layout ..................................................................................................... 115
Fig. 4.25 Position 1 ................................................................................................................................... 116
Fig. 4.26 Position 2 ................................................................................................................................... 116
Fig. 4.27 DM noise measurement result ................................................................................................... 117
Fig. 4.28 Insertion voltage gain for filter with M1,2 cancellation .............................................................. 117
Fig. 4.29 Circuit network 1 ....................................................................................................................... 118
xvii
Fig. 4.30 Circuit network 2 ....................................................................................................................... 119
Fig. 4.31 ESL cancellation of capacitor .................................................................................................... 119
Fig. 4.32 ESL cancellation implementation into PCB .............................................................................. 120
Fig. 4.33 Calculation and simulation of PCB winding inductance ........................................................... 120
Fig. 4.34 Insertion voltage gain for filter with ESL cancellation .............................................................. 121
Fig. 4.35 DM noise measurement result with ESL cancellation ............................................................... 122
Fig. 4.36 Wound technology ..................................................................................................................... 122
Fig. 4.37 Stacked-film technology ............................................................................................................ 123
Fig. 4.38 Current flow path in the film capacitor ...................................................................................... 123
Fig. 4.39 Flux linkage between two capacitors ......................................................................................... 124
Fig. 4.40 Cancellation of M3 using additional inductor ............................................................................ 124
Fig. 4.41 Equivalent circuit of M3 cancellation method ........................................................................... 125
Fig. 4.42 Insertion voltage gain of filter with M3 cancellation ................................................................. 125
Fig. 4.43 DM noise measurement result with M3 cancellation ................................................................. 126
Fig. 4.44 DM filter performance ............................................................................................................... 127
Fig. 4.45 CM filter performance ............................................................................................................... 127
xviii
List of Tables
Table 3-1 Loss breakdown for non-coupled inductor ................................................................................. 56
Table 3-2 Loss breakdown for coupled inductor Structure I ...................................................................... 57
Table 3-3 Loss breakdown for coupled inductor Structure II ..................................................................... 58
Table 3-4 Loss breakdown for coupled inductor Structure III .................................................................... 59
Table 3-5 Loss breakdown for litz wire inductor ........................................................................................ 84
Table 3-6 Loss breakdown for conventional PCB winding inductor .......................................................... 85
Table 3-7 Loss breakdown for proposed PCB winding inductor with interleave ....................................... 86
Table 3-8 Loss breakdown for optimized inductor design .......................................................................... 90
Table 4-1 Comparison between calculation and FEA simulation ............................................................. 121
Yuchen Yang Chapter 1
1
Chapter 1. Introduction
1.1 Introduction to EMI and EMC
In a world full of electronic devices, it is very important that these devices can work properly
in a complicated electromagnetic environment. Thus, electromagnetic compatibility (EMC) is a
significant characteristic of electronic devices. The definition of EMC according to International
Electrotechnical Commission (IEC) is “The ability of an equipment or system to function
satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic
disturbances to anything in that environment.”[1]. Nowadays, EMC is becoming more and more
important. First, more and more applications are going to be electrified, such as automobile, ship,
and aircraft. Every devices in these applications must work properly to assure safety. In addition,
recent years have seen a dramatically growth of personal electronic devices. We are facing a more
complicated electromagnetic environment that we have never met before. Second, modern
electronic devices are more susceptible to electromagnetic disturbances. The lower supply voltage
is more easily affected by noises. Higher switching frequency is introducing severe disturbances.
Furthermore, the pursuing of power density is making electronic devices closer to each other.
Electromagnetic interference is defined as “degradation of the performance of an equipment,
transmission channel or system caused by an electromagnetic disturbance” [2]. There are four
methods that the electromagnetic interference can propagate from the noise source to the victim:
(1) Conducted interference;
(2) Radiated interference;
Yuchen Yang Chapter 1
2
(3) Inductive interference;
(4) Capacitive interference.
The conducted interference means the noise propagates through conductors. Radiated
interference propagates through the radiation of electromagnetic field. The inductive and
capacitive interference combined as the near field coupling propagates from the noise source to
the victim.
EMI can cause malfunction of electronic devices, from the annoying cell phone interference
to lethal failure of life support equipment in the hospital. In order to avoid those impacts, there are
several EMI standards to limit both the conducted and radiated EMI noise, such as FCC part 15 in
United States and CISPR 22 in Europe. Fig. 1.1 shows the EN55022 Class B standards for
conducted EMI noise.
Fig. 1.1 EMI standard EN55022 Class B for conducted noise
Yuchen Yang Chapter 1
3
1.2 EMI Measurement
Because of the complexity of EMI propagation path, different measurement setup leads to
different result. In order to get a constant measurement condition, EMI standards provide the
specified measurement setup. The FCC measurement setup is shown in Fig. 1.2.
Fig. 1.2 FCC measurement setup
Line Impedance Stabilization Network (LISN) is used to guarantee constant measurement
condition. Fig. 1.3 shows the internal circuit of a LISN. The main function of LISN is that it
provide a stable loop impedance that can get repeatable measurements of the EMI noise. LISN
works like a low-pass filter that it can block high frequency noise from the power source and allow
low frequency power flow to the Equipment-Under-Test (EUT).
Yuchen Yang Chapter 1
4
In the real measurement setup, two LISNs are needed to connect between power source and
EUT as shown in Fig. 1.4. A spectrum analyzer is connected to one LISN to pick up the total noise
while a standard 50Ω terminator is connected to the other LISN.
Fig. 1.3 Internal circuit of a LISN
Fig. 1.4 Practical noise measurement circuit
There are three different methods to measure EMI noise: peak, average and quasi-peak. The
peak detector picks the maximum amplitude of the noise signal for each frequency. The average
detector takes an envelope-detected signal and passes it through a low-pass filter with a bandwidth
1uF
50uH
0.1uF
1kΩ
AC in AC out
To 50Ω terminator or spectrum analyzer
Grounded Shield
LISN
50uH
0.1uF
1kΩ
AC in
LISNs
0.1uF
1kΩ
50uH
1uF×2
50Ω Terminator
50Ω Spectrum Analyzer
Equipment Under Test~
Yuchen Yang Chapter 1
5
much less than the resolution bandwidth. The filter integrates (averages) the higher-frequency
components such as noise. Quasi-peak detector is a weighted form of peak detection. The
measured value of the quasi-peak detector drops as the repetition rate of the measured signal
decreases [3].
1.3 Conducted EMI Noise in Switch Mode Power Supplies
Switch mode power supplies are widely used in different applications, such as power supply
for telecom and server, silver box for desktops, adapter for laptops and charger for personal mobile
devices.
Switch mode power supplies is a complicated system and have different structures for
different applications. Fig. 1.5 shows the structure of power supply below 75W. Because the power
is below 75W, it do not need a power factor correction (PFC) circuit. The diode bridge rectifies
the AC input voltage to DC voltage. The isolated DC-DC converter converts the DC voltage to
desired DC bus voltage. Then the voltage regulator (VR) module converts the bus voltage to
different load voltage and assures the required transient performance.
Fig. 1.5 Structure of low power converter
For the power supply for server and telecom system, the structure is a little different. Fig. 1.6
shows the structure of front-end converter for server and telecom system. First, the PFC converter
rectifies the AC input voltage to 380V DC voltage. Second, the DC-DC converter converts the
380V DC voltage to a 48V/12V DC bus voltage. Then the voltage regulator module will take over.
Yuchen Yang Chapter 1
6
Fig. 1.6 Front-end converter for server and telecom
Switch mode power supplies never stop pursuing high efficiency and high power density. In
order to reach this goal, new semiconductor devices, such as GaN and SiC, are being developed.
In addition, circuit topologies, structure and control method continue improving to help make
switch mode power supplies smaller, lighter and more efficient.
However, high dv/dt and di/dt in switch mode power supplies will cause severe EMI noise
issue. Hence, EMI filter is always needed in switch mode power supply to attenuate severe
conducted EMI noise. Fig. 1.7 shows a state-of-the-art front-end converter for server system. It
can be seen that the EMI filter takes about 1/4 to 1/3 volume of total converter. Reducing the noise
of power converter can reduce the size and volume of EMI filter and help improving the power
density of total converter.
Fig. 1.7 State-of-the-art front end converter
The conducted EMI noise is always decoupled to differential mode (DM) noise and common
mode (CM) noise. Because the DM noise and CM noise have different propagation path, they need
Yuchen Yang Chapter 1
7
different attenuation method. Fig. 1.8 shows the propagation path of DM conducted noise. DM
noise current iDM flows between two power lines. Fig. 1.9 shows the propagation path of CM
conducted noise. CM noise current iCM flows between power lines and ground.
Fig. 1.8 DM noise propagation path
Fig. 1.9 CM noise propagation path
According to the EMI standard EN55022, the noise voltage on the 50Ω resistor is limited:
{𝑣1 = 50(𝑖𝐷𝑀 − 𝑖𝐶𝑀)𝑣2 = 50(−𝑖𝐷𝑀 − 𝑖𝐶𝑀)
(1-1)
v1 and v2 both can represent total EMI noise. In the measurement, the amplitudes of v1 and v2
are almost identical because iDM and iCM are high frequency AC currents [5].
Yuchen Yang Chapter 1
8
DM noise voltage and CM noise voltage can then be represented as:
{𝑣𝐷𝑀 = 50𝑖𝐷𝑀 =
𝑣1−𝑣2
2
𝑣𝐶𝑀 = −50𝑖𝐶𝑀 =𝑣1+𝑣2
2
(1-2)
For off-line switch mode power supplies, DM noise is dominated by PFC converter. CM noise
is a more complicated issue. It is contributed by both PFC converter and DC/DC converter. A lot
researches have been done to reduce the CM noise of PFC converter [6]-[11]. However, the CM
noise of DC/DC converter still remains a challenge.
1.4 EMI Noise Reduction Methods for Switch Mode Power Supplies
There are three types of CM noise reduction method, symmetry, balance and shielding.
CM noise current is formed by severe dv/dt on parasitic capacitance, as shown in (1-3). The
idea of symmetry technique is forming two dv/dt sources that have same amplitude but opposite
direction. With same capacitance in the propagation path, the two noise sources can cancel each
other and minimize CM noise. Before the balance technique was proposed, several methods had
been provided to reduce CM noise by using symmetry concept [12]-[19].
𝑖𝐶𝑀 = 𝐶𝑑𝑣
𝑑𝑡 (1-3)
Fig. 1.10 shows the topology of full bridge LLC converter. The two switch nodes on primary
side are symmetrical. When V1 has a positive dv/dt, V2 will have a negative dv/dt with same
amplitude. The two switch nodes on the secondary side are also complimentary. Therefore, the
CM noise generated by these noise sources will be canceled by each other.
Yuchen Yang Chapter 1
9
Fig. 1.10 Full bridge converter
However, in the real circuit, it is very difficult to match the two complimentary dv/dt sources.
As shown in Fig. 1.11, the dv/dt of V2 is a little slower than V1. Thus, the two noise sources cannot
be completely cancelled. And CM noise is not minimized.
Fig. 1.11 Waveforms of V1 and V2
Another issue for symmetry method is that most circuit topologies are asymmetrical in nature.
In order to achieve symmetry, additional components is required to be added in the circuit, such
as diodes, power switches and windings. In [19], CM noise is cancelled by achieving symmetrical
circuit topology for boost converter. Fig. 1.12 demonstrates the symmetrical boost topology versus
a conventional boost topology [19]. The original boost inductor L is split into two identical coupled
inductor L1 and L2. In addition, another diode D2 is introduced. Thus, another dv/dt source is
introduced and it has the same amplitude and opposite polarity with node A. The parasitic capacitor
Yuchen Yang Chapter 1
10
Cd1 and Cd2 is controlled to be the same value. Hence, the CM noise current flowing through Cd1
and Cd2 has the same amplitude but out-of-phase. Consequently, there is no net CM noise current
flowing to the ground and CM noise has been cancelled.
(a) Conventional boost converter
(b) Symmetrical boost converter
Fig. 1.12 Boost converter with symmetry technique
Although the experiment shows significant CM noise reduction, symmetry technique has its
own limitation. The additional components will make the converter more complicated and
introduces extra loss and higher cost. It sacrifices too much to reduce CM noise, which is not
applicable in commercial products.
L
S
D
CL RL VO
+
-
Cd
Vin
L1
S
D1
CL RL VO
+
-
Cd1
L2
D2Cd2
Vin
Yuchen Yang Chapter 1
11
In order to have good CM noise reduction without extra loss, balance technique is provided
[7]. As shown in Fig. 1.13, the inductor is split in to L1 and L2, but there is no need to add another
diode into the converter. Cd is the parasitic capacitance between MOSFET drain and ground. Ca is
the parasitic capacitance between the cathode of diode D, large area of output traces, load and the
ground. Cc is the parasitic capacitance between load, large area of output traces and the ground.
Fig. 1.13 Balanced boost converter
According to the substitution theorem, if the MOSFET branch is substituted by a voltage
source V having the same voltage as the original branch, the circuit behavior keeps the same.
Hence, the MOSFET can be replaced by a voltage source V, as shown in Fig. 1.14.
Yuchen Yang Chapter 1
12
Fig. 1.14 Replace the MOSFET with a voltage source
In the CM noise frequency range, the input and output capacitor can be treated as short circuit.
In addition, the diode bridge can also be treated as short circuit since it is always conducting current.
Thus, the CM noise model of this boost converter is shown in Fig. 1.15.
Fig. 1.15 CM noise model of balanced boost PFC converter
It can be seen that this CM noise model is a Wheatstone bridge circuit. In a general Wheatstone
bridge structure shown in Fig. 1.16, if the balance equation (1-4) is satisfied, point B and point D
have identical voltage and there is no current going outside this circuit.
Yuchen Yang Chapter 1
13
Fig. 1.16 Wheatstone bridge structure
𝑍1
𝑍2=
𝑍3
𝑍4 (1-4)
Thus, for the CM noise model of boost PFC converter in Fig. 1.15, as long as the balance
equation (1-5) is satisfied, there is no CM noise current generated by converter going through
LISN.
𝑍𝐿1
𝑍𝐿2=
𝑍𝐶𝑑
𝑍𝐶𝑎+𝐶𝑐 (1-5)
In contrast to the symmetry technique, this ratio in the balance technique do not need to be
only 1:1. The impedance ratio can be any value as long as the balance condition is satisfied, CM
noise can be cancelled. Hence, the balance technique offers additional freedom to the designer to
choose the impedance ratio according to the converter design. Furthermore, balance technique do
not require additional components in the circuit. This will not increase the complexity of the
converter and will not increase the cost and loss while has significant CM noise reduction, which
is preferred in the commercial products.
VAC
Z1
Z2
A
C
Z4
Z3
B D
Yuchen Yang Chapter 1
14
However, the balance equation is only valid at low frequency. At high frequency, the
equivalent parallel capacitor (EPC) and equivalent parallel resistor (EPR) of the inductor dominant
the impedance of inductor branch. Therefore, at high frequency, the balance equation may not be
achieved. Fig. 1.17 [8] shows the CM noise reduction result with balance technique for boost
converter. It can be seen that at low frequency, the balance technique can reduce CM noise. But
after 2 MHz, the CM noise is not reduced.
Fig. 1.17 CM noise reduction of boost converter with balance condition achieved at low
frequency [8]
In order to improve the high frequency performance of balance technique, two inductors are
coupled together, as shown in Fig. 1.18 [8]. It can be seen that with coupling, the ratio of
inductance, EPC and EPR are all turns ratio of two inductor windings. The balance condition can
be simplified as (1-6).
Yuchen Yang Chapter 1
15
(a) CM noise model (b) Equivalent decoupled circuit
Fig. 1.18 Balance technique with coupled inductors [8]
𝑛 =𝐶𝑏𝐶𝑑
(1-6)
The CM noise reduction result is shown in Fig. 1.19 [8]. The CM noise can be reduced
significantly. However, the CM noise reduction is not good beyond 8 MHz. It is because that the
balance equation (1-6) has an assumption: the coupling coefficient between L1 and L2 is 1.
However, for a litz wire inductor with toroid core, it is difficult to achieve high coupling coefficient.
The coupling coefficient between L1 and L2 are usually below 0.8. Therefore, the high frequency
performance of balance technique is not as good as expected.
Fig. 1.19 CM noise of balanced boost converter with inductor coupling [8]
Yuchen Yang Chapter 1
16
For the isolated DC/DC converter, the main CM noise path is the transformer. Take flyback
converter as an example, it has two switches and one transformer, as shown in Fig. 1.20. The two
switches can be modeled as two noise sources. The parasitic capacitance between the primary
winding of transformer and secondary winding of transformer are the main propagation path of
CM noise. Therefore, shielding method is very popular to reduce the CM noise in isolated
converters.
Fig. 1.20 Flyback converter and CM noise path
In a practical transformer, there is a large amount displacement current flowing between the
primary and secondary windings. This contributes large CM noise current. The purpose of
shielding is to block the CM noise propagation path through the transformer. [20] provides a
method to block the CM noise current through transformer. Fig. 1.21 shows the proposed
transformer structure. In this figure, half core of the transformer is implemented into the circuit.
The primary winding has three layers and secondary winding has one layer. Instead of using one
complete wire as the primary winding, this method breaks the primary winding into two parts. One
part is normally winded primary winding P1, P2. The other part P3 is a shielding winding that is
identical with secondary winding but connected in series with the primary winding. The
RLISN
+-
NP NS
Adv
dt
-+
Ddv
dt
A
B D
C
Yuchen Yang Chapter 1
17
displacement current flows between primary windings does not contribute to the CM noise current
because this current will circulating within the primary side. Fig. 1.22 shows the voltage
distribution of the shielding and secondary windings. Assuming the voltage distributes linearly
along the winding. It can be seen that for a specific point on the primary winding, it shares the
same voltage potential as the corresponding point on the secondary winding. Consequently, there
is no voltage difference between the shielding and secondary winding and no displacement current
can flow between the shielding and secondary windings. Thus the CM noise current flowing
through the transformer is blocked.
Fig. 1.21 Series Shielding method
Yuchen Yang Chapter 1
18
Fig. 1.22 Voltage distribution of P3 and S
However, this method requires that the shielding and secondary winding must be aligned
perfectly. Fig. 1.23shows that if there is a misalignment between shielding and secondary winding,
the voltage distribution will change. As shown in Fig. 1.23 (b), there will be voltage difference
between the shielding and secondary winding and this will lead to CM noise current flow through
the transformer. The benefit of this technique will be diminished. Because of this, this method is
not effective in mass production.
(a) Winding position when misalignment (b) Voltage distribution of misalignment windings
Fig. 1.23 Misalignment between primary and secondary windings
Yuchen Yang Chapter 1
19
In [21], a shielding method is provided that can reduce CM noise using cancellation concept.
Fig. 1.24 shows its transformer structure. Instead of using a shielding that blocking the whole
winding area, it reduces the shielding width and leaves a non-shielded area. In non-shielded area,
the voltage of primary winding is higher than secondary winding. Displacement current i1 flows
from primary to secondary. In the shielded area, the voltage of secondary winding is higher than
shielding, so the displacement current i2 flows from secondary to primary. The equivalent CM
noise model is shown in Fig. 1.25. The parasitic capacitor between primary winding and secondary
winding forms the capacitor CAC. The parasitic capacitor between secondary winding and shielding
forms the capacitor CBD. In order to achieve i1 = i2, (1-7) should be achieved. The value of CAC
and CBD can be adjusted by changing the length of shielding.
Fig. 1.24 Partial shielding method
Yuchen Yang Chapter 1
20
Fig. 1.25 CM noise model of partial shielding
𝐶𝐴𝐶𝑑𝑉𝐴𝑑𝑡
= 𝐶𝐵𝐷𝑑𝑉𝐷𝑑𝑡
(1-7)
However, this method has two issues. The first one is that the value of dVA/dt and dVD/dt are
very different. The two noise sources are from two different devices, as shown in Fig. 1.26. It is
very difficult to match dVA/dt and dVD/dt with exact ratio. In the real production, the optimal width
of shielding is obtained by trial-and-error procedure and requires a lot effort. The second issue is
that the value of CAC and CBD is very difficult to calculate and control in mass production. Hence
this method requires a precise control on the relative position of windings and shielding. However,
it is hard to guarantee this for every transformer in the real production and will diminish the effect
on the CM noise reduction.
Fig. 1.26 Waveform of VA and VD
Yuchen Yang Chapter 1
21
1.5 Proposed Dissertation Outline
As discussed in this chapter, the EMI is an important issue in switch mode power supply
design. With the pursuing of higher efficiency and higher power density, new method to help
reduce EMI noise is needed. With the wide-bandgap devices, the switching frequency of power
converters has been increased 10 times higher. The design of power converter has been
dramatically changed. Therefore, the EMI analysis needs to be viewed in a new angle. High
switching frequency provides both challenges and opportunities in EMI research.
Chapter 2 introduces the shielding technique for PCB winding transformer. With wide
bandgap (WBG) devices, the switching frequency of converters has been increased 10 times higher
than traditional Si MOSFET design. This dramatically changes the magnetic design. With the PCB
winding transformer, the converter is smaller and efficient. However, due to large inter-winding
capacitance, the CM noise of PCB winding transformer is very large. A novel shielding method
for PCB winding transformer is proposed. This shielding method can not only effectively reduce
the CM noise, but also improve the converter efficiency. Furthermore, the PCB shielding method
can also be implemented into matrix transformer.
Chapter 3 proposes the novel PCB winding inductor for CRM PFC converter. In today’s
industry practice, litz wire inductor is widely used because of high efficiency. With the increasing
demanding of high efficiency and high power density, the litz wire inductor is not suitable. In this
chapter, two PCB winding inductor designs are proposed, one for 1 kW server power supply,
another for 6.6 kW on-board battery charger of electric vehicle. The challenge of PCB winding
inductor design is high AC winding loss. In this chapter, a novel winding structure is proposed to
solve this issue. The PCB winding inductor can achieve similar efficiency as litz wire inductor
Yuchen Yang Chapter 1
22
with smaller volume and reduced labor intensive work. Furthermore, the balance technique is
applied to PCB winding inductor to help reduce CM noise. With PCB winding inductor, the
balance technique has better high frequency performance.
Chapter 4 investigates the EMI filter design for 1 kW server power supply. Conventionally,
people have to use two-stage EMI filter to meet the EMI standard. The two-stage EMI filter are
bulky and expensive. With the EMI noise reduction methods proposed in previous chapter, one-
stage EMI filter can be applied. This chapter analyzes the self-parasitic and mutual coupling of
one-stage EMI filter. The near field measurement method is applied to visualize the flux near filter
components. The methods to reduce the self parasitics and mutual coupling of one-stage EMI filter
is analyzed and demonstrated in 1 kW server power supply system.
Chapter 5 gives conclusions and discussion about future work.
Yuchen Yang Chapter 2
23
Chapter 2. Shielding Technique for Isolated DC/DC
Converter
2.1 PCB Winding Transformer with Shielding Technique
People always want the power supply to be smaller and more efficient. That makes the power
density and efficiency are the two major driving forces for power supply system [22][23]. With
traditional Si MOSFETs, the switching frequency of power converter is limited to 100 kHz. With
low switching frequency, the magnetic components are usually bulky. In addition, because of the
high volt-sec, the magnetic components have to use many turns of litz wire. The relationship
between core size and voltage-sec can be expressed as:
𝑁 ∙ 𝐴𝑒 =𝑉𝑂(1−𝐷)
2∙∆𝐵∙𝑓𝑠 (2-1)
For conventional Si based flyback converter, the turns number of transformer is very high. As
shown in Fig. 2.1, the primary winding has 39 turns, and the secondary winding has 7 turns. People
use litz wire to build this transformer. It is impossible to build such many turns with PCB winding
and maintaining high efficiency.
Yuchen Yang Chapter 2
24
Fig. 2.1 Conventional flyback converter transformer
However, with the development of wide-band-gap power devices, including gallium nitride
(GaN) transistors and silicon carbide (SiC) transistors, the switching frequency of power converter
can be pushed to hundreds of kHz, even above MHz. At such high switching frequency, the size
of magnetic components can be reduced a lot. Furthermore, the turns number can also be greatly
reduced. This provides an opportunity for magnetic components to use PCB winding. Comparing
with traditional litz wire magnetics, the PCB winding magnetics can achieve small volume, better
tolerance control, and automated production without labor intensive work. Therefore, the PCB
winding is a promising technique for high frequency converters.
The 65W flyback adapter discussed in previous section can be built with GaN devices
switching above 1MHz [24]. In this flyback converter, PCB winding transformer is applied in the
4-layer PCB motherboard. As shown in Fig. 2.2, the 1 MHz PCB winding transformer is 5 times
smaller than traditional litz wire transformer.
Yuchen Yang Chapter 2
25
Fig. 2.2 Size comparison between 1 MHz PCB winding transformer and 100 kHz litz wire
transformer
Fig. 2.3 shows the cross-section view of PCB winding transformer. The primary winding has
10 turns and secondary winding has 2 turns. It has much less turns than the traditional low
frequency litz wire transformer. Thus, the transformer can be built with 4-layer PCB motherboard.
Fig. 2.3 Cross-section view of PCB winding transformer
However, the PCB winding transformer has a significant drawback, which is very large
parasitic capacitance between primary winding and secondary winding. Large capacitance will
lead to large CM noise current. Fig. 2.4 shows the CM noise measurement result for the PCB
winding flyback converter. The noise amplitude is much higher than low frequency flyback
converter with litz wire transformer. Furthermore, the CM noise remains very large on high
frequency range. This brings a great challenge to EMI filter design.
Yuchen Yang Chapter 2
26
Fig. 2.4 CM noise measurement for PCB winding flyback converter
In order to solve this issue, a novel shielding technique is applied to PCB winding transformer
[25]. Fig. 2.5 shows the circuit topology and transformer cross-section view of flyback converter
with shielding layer. The shielding is connected to primary side. Therefore the CM noise current
generated by primary side will circulate within primary side. If the shielding is made exactly same
and align with secondary winding, as shown in Fig. 2.6, there will be no CM noise current between
shielding and secondary winding. Thus, the CM noise is blocked from primary winding to
secondary winding. However, in the real implementation, there must be a jumper wire to connect
the shielding to primary winding. In a high frequency transformer design, the jumper wire is not
acceptable. Therefore the shielding layer must have a different layout design.
Yuchen Yang Chapter 2
27
(a) Flyback converter with shielding
(b) Transformer cross-section view
Fig. 2.5 Flyback converter with shielding layer
Fig. 2.6 Detailed PCB layout with jumper connection
Fig. 2.7 shows that shielding has 180º rotation with secondary winding. Thus, it is easy to
connect the shielding with primary ground. However, after rotation, there will be voltage potential
Yuchen Yang Chapter 2
28
difference between shielding and secondary winding. Fig. 2.8 shows 3D structure of shielding and
secondary winding. Point B, C, E and F are marked in Fig. 2.5(a), and point F on shielding layer
is connected to primary ground. In order to analyze the displacement current flowing between
shielding layer and secondary winding, the voltage distribution needs to be described. Shielding
layer can be treated as a one-turn winding. There will be voltage induced on the shielding layer
according to Faraday’s Law. An X-Y coordinate system is built to analyze the voltage distribution
of shielding layer and secondary winding. Zero point is set between point F and point E on
shielding. The X axis is built along the winding and representing the position of certain point on
the winding. The Y axis indicates the voltage potential of certain point. Fig. 2.9 shows the voltage
distribution of shielding layer and secondary winding. One assumption is made that the voltage is
linearly distributed along the winding. For shielding, from point E to point F the voltage decreases
from V to 0. For secondary winding, because the zero point is set in the middle of secondary
winding, at zero point, the voltage on shielding is V/2. First, the voltage on shielding decreases
linearly from V/2 to 0. Then, at point B, the voltage is V and decreases to V/2. It can be seen that
on one half, the voltage of secondary winding is higher than shielding. The displacement current
flows from secondary winding to shielding. On the other half, the voltage of shielding is higher
than secondary winding. The displacement current flows from shielding to secondary winding.
Those two currents have same amplitude and opposite direction. Thus, there is no net CM noise
current flowing out of the transformer. In summary, the function of shielding is: (1) it blocks the
displacement current from primary side; (2) the displacement current between shielding and
secondary winding will not generate CM noise current.
Yuchen Yang Chapter 2
29
Fig. 2.7 Shielding has a 180º rotation with secondary winding
Fig. 2.8 3-D structure of shielding and secondary winding
Fig. 2.9 Voltage distribution of shielding and secondary winding
Furthermore, the shielding can be integrated into primary winding, so that the shielding serves
as the first two turns of primary winding, as shown in Fig. 2.10. The transformer cross-section
Yuchen Yang Chapter 2
30
view is shown in Fig. 2.11. With the integration, the transformer can achieve 13% winding loss
reduction as well as shielding effect.
Fig. 2.10 Shielding serving as part of primary winding
Fig. 2.11 Transformer cross-section view with shielding
Fig. 2.12 shows the CM noise measurement result for the PCB winding flyback converter with
shielding. It can be seen that shielding can achieve 27dB noise reduction. Furthermore, shielding
remains very effective on high frequency up to 30 MHz. Since high frequency noise attenuation is
the most difficult part on designing EMI filter. This shielding method can greatly benefit EMI
filter design.
Yuchen Yang Chapter 2
31
Fig. 2.12 CM noise reduction for flyback converter with shielding
2.2 Matrix Transformer with Shielding Technique
Pulse-width modulation (PWM) converters is widely used in front-end DC/DC converters.
However, because of the hard switching performance, the switching loss is large in PWM
converters. This limits the efficiency of PWM converters and impedes the converter going to high
switching frequency. Hence, it is hard for PWM converter to achieve high power density and high
efficiency. Soft switching technique can make PWM converter achieve zero voltage switching
(ZVS). So the switching loss of PWM converter can be reduced and it has the ability to go to high
switching frequency to improve power density. However, another disadvantage of PWM converter
still limits its performance. For the computing electronic systems, such as desktop, laptop, server
and telecom applications, certain hold-up time is required to assure data security. Normal PWM
converters require bulky hold up capacitors to guarantee the hold-up time. Thus the power density
of PWM converters are still limited.
Yuchen Yang Chapter 2
32
LLC resonant converter, is becoming more and more popular because of the high efficiency
and high power density [24]-[32]. LLC resonant converter can achieve ZVS for primary side
switches from zero to full load range. Meanwhile, the secondary side synchronous rectifier (SR)
can achieve zero current switching (ZCS). Because of this, the switching loss of LLC resonant
converter can be minimized and high efficiency can be achieved. In addition, low switching loss
provides the LLC resonant converter the ability to achieve high switching frequency. This can
continue reduce the size of passive components and improve power density. Furthermore, it is easy
to integrate magnetic components into the transformer to further reduce size and cost of LLC
resonant converter. In addition, the voltage gain characteristic of LLC resonant converter makes it
have good hold-up capability to reduce the volume of hold-up capacitor. In summary, LLC
resonant converter is becoming dominant in off line power supply systems.
The server applications require low output voltage and high output current. For example, a 1
kW, 12V output server power supply will have more than 80 A output current. 4-8 SRs should be
paralleled to reduce the secondary conduction loss. The circuit is shown in Fig. 2.13 (a). This
design will have several issues. First, it is difficult to achieve both static and transient current
sharing between the paralleled SRs. Second, there will be large AC current flow through the
terminations of transformer and SRs (red dots in Fig. 2.13(a)), which will generate large
termination loss. Third, because of the low switching frequency, the transformer has to use litz
wires and copper foil to build primary and secondary winding. In this way, the transformer is bulky
and expensive. The manufacture procedure is labor intensive and the quality control is poor. Fig.
2.13 (b) shows the prototype of the conventional LLC converter.
Yuchen Yang Chapter 2
33
(a) Circuit schematic (b) Hardware prototype
Fig. 2.13 Conventional LLC converter
The matrix transformer is defined as an array of elements intertwined so that the whole
functions as a single transformer [33][34]. Each element being a single transformer that contains
a set turns ratio, i.e. 1:1, 2:1 ....n:1. The desired turns ratio is obtained by connecting the primary
windings of the elements in series and the secondary’s in parallel. For the high current design cases
only a single turn secondary will be considered. The benefits of the matrix transformer are that it
can split current between secondary windings connected in parallel, reduce leakage inductance by
lowering the N2 value of the secondary loop inductance, and improve thermal performance by
distributing the power loss throughout the elements [35]. In addition, the matrix transformer
structure also can effectively reduce the magnetomotive force (MMF) of the windings, especially
for the PCB winding. That also means a reduction in leakage inductance and winding AC
resistance [36]. This makes the matrix transformer a very attractive structure for high current, high
switching frequency applications.
Fig. 2.14 shows a design of 400V/12V, 1kW LLC resonant converter with matrix transformer
[36]. The switching frequency of this converter is 1MHz. It uses GaN devices as primary switches
to reduce switching related loss [37][38]. By utilizing flux cancellation method [39], this converter
Yuchen Yang Chapter 2
34
uses two U-I cores for four sets of transformers. Comparing with normal design, this can reduce
the core loss and core size by more than 30% [36]. For each small cell of transformer, the primary
winding has four turns and secondary winding has two one-turn center-tap structure. The primary
windings are in series and secondary windings are in parallel. Hence, the total turns ratio is 16:1.
As shown in Fig. 2.15, by using matrix transformer structure, four-layer PCB can achieve 16:1
turns ratio. Primary windings are at the two middle layers and secondary windings are at top and
bottom layers. Thus, secondary SR devices and output capacitors can be directed mounted on top
of the secondary winding, as shown in Fig. 2.16. By doing this, there is no extra termination loss.
The primary and secondary windings are exactly overlapped.
This LLC resonant converter can achieve more than 95% efficiency and 710W/in3 power
density, which is comparable with the state-of-the-art industry product.
Fig. 2.14 LLC resonant with matrix transformer
Yuchen Yang Chapter 2
35
Fig. 2.15 Matrix transformer structure
Fig. 2.16 Mount secondary devices and output capacitors on top of secondary winding
Matrix transformer brings a lot of benefits. However, there is always a price to pay. In the
matrix transformer, the winding area is large, which can introduce large inter-winding capacitance.
Furthermore, because of the interleaving structure, the inter-winding capacitance is further
increased. This large inter-winding capacitance between primary and secondary windings will
bring large CM noise current. As shown in Fig. 2.17, the measured CM noise of LLC converter is
way above the EMI standard. Large CM noise filter is needed to achieve the required EMI noise
level. The high power density may be diminished by huge CM noise filter.
Yuchen Yang Chapter 2
36
Fig. 2.17 CM noise measurement result for LLC resonant converter with matrix transformer
The shielding method is provided for matrix transformer to reduce CM noise [40]. Fig. 2.18
shows the circuit structure after adding shielding layers between primary and secondary windings.
Fig. 2.19 shows the cross section view of transformer with shielding layers. Shielding layer is
added between each primary-secondary cell. Thus, the transformer need six layers of PCB after
adding shielding. It can be seen from Fig. 2.18 that all shieldings are connected to the primary
ground. Thus, the noise current coming from primary side will be blocked by shielding and
confined in the primary side. This noise current will not cause CM noise.
Yuchen Yang Chapter 2
37
Fig. 2.18 Adding shielding into matrix transformer
Fig. 2.19 Transformer structure with shielding
However shielding, treated as a one-turn winding, becomes another voltage pulse source.
Therefore the shielding need to be carefully designed otherwise there will be CM noise current
between shielding and secondary winding. One shielding-Sec. winding is used as an example to
illustrate the shielding design. As shown in Fig. 2.20, Secondary winding is on top of the shielding,
and node A, B, A’, B’ is noted in Fig. 2.18. Shielding is rotated 180 degree with secondary winding
to have an easy connection to primary ground. Assuming voltage is distributed linearly along the
winding, Fig. 2.21 Voltage distribution of shielding and secondary winding shows the voltage
Yuchen Yang Chapter 2
38
distribution of shielding and sec. winding. The coordinate is build: set zero on point A and the x-
axis is along the winding; y-axis is the voltage of each point on the winding. For secondary winding,
node A is the dv/dt point and node B is connected to the ground. Hence from A to B, the voltage
distributes from V to 0. For shielding, node B’ is connected to ground, and the zero point of the
axis is at the middle point of shielding. Hence the voltage of shielding first distributes from V/2 to
0, then jump to V (node A’) and decrease to V/2. It can be seen that on one half, the voltage of
secondary winding is higher than shielding. There is displacement current go from secondary
winding to shielding. On the other half, the voltage of shielding is higher than secondary winding.
There is displacement current go from shielding to secondary winding. These two current are at
the same magnitude but has opposite direction. Therefore, displacement current is circulating
within shielding and secondary winding. Therefore, the shielding is designed to not only block the
primary noise source, but also block the noise from secondary side.
Fig. 2.20 3D structure of secondary winding and shielding
Yuchen Yang Chapter 2
39
Fig. 2.21 Voltage distribution of shielding and secondary winding
The experiment result is shown in Fig. 2.22. For an easy observation, the peak points are
connected as the envelope. It can be seen that the proposed shielding technique can have a 23dB
reduction at 1MHz. Further, the shielding is still effective at very high frequency range.
Fig. 2.22 CM noise measurement result for matrix transformer with shielding
Yuchen Yang Chapter 2
40
The efficiency comparison is shown in Fig. 2.23. It can be seen that less than 0.5% efficiency
difference between shielding version and original design. In summary, the shielding technique has
good CM noise reduction result while little sacrifice of efficiency.
Fig. 2.23 Efficiency comparison between shielding version and original version
2.3 Conclusions
In this chapter, the shielding method is proposed to reduce CM noise for isolated converters.
With GaN and SiC devices, the switching frequency of power converters has been increased 10
times higher. This provides the opportunity to use the PCB winding transformer. The PCB winding
transformer is high efficient, high power density and good manufacturability. However, the CM
noise of PCB winding transformer is very high due to large inter-winding capacitance. In this
chapter, a novel shielding method is proposed to reduce the CM noise of PCB winding
transformers. The implementation of this shielding method is simple. The shielding method can
effectively reduce the CM noise of a 65 W flyback converter by 27 dB. Furthermore, the shielding
Yuchen Yang Chapter 2
41
remains effective up to 30 MHz. In addition, the shielding can be integrated with primary winding
to further reduce the winding loss of the transformer.
The PCB shielding method can also be applied to matrix transformer structure. A 1 kW LLC
converter with matrix transformer is set as an example. With this shielding method, the CM noise
can be reduced by 23 dB. The shielding has 10 dB to 20 dB attenuation at very high frequency.
This can greatly help the EMI filter design since the high frequency noise is the most difficult to
attenuated by EMI filter.
Yuchen Yang Chapter 3
42
Chapter 3. Balance Technique for Coupled Inductors
Integrated into PCB
In previous chapter, the magnetic integration and EMI noise attenuation for DC/DC converter
is discussed. For the power over 75W, the PFC converter is required in the power supply to ensure
good power factor. In commercial products, the switching frequency of the PFC converter is
usually lower than 100 kHz. Fig. 3.1 shows a typical 1 kW AC-DC converter. With low switching
frequency, the converter is very bulky. Over 40% of the components are manually assembled,
especially the inductors and transformers. Manufacturing this converter is labor intensive.
Furthermore, the tolerance control of these components is poor. The PFC converter consumes
approximately one third of the power supply. Increasing the switching frequency can reduce the
volume of the PFC converter. Furthermore, increasing the switching frequency can greatly raise
the corner frequency of the EMI filter and reduce the filter size. Thus, the power density of power
supply can be increased significantly.
Fig. 3.1 A typical AC-DC converter
Yuchen Yang Chapter 3
43
With wide-band-gap devices, the switching frequency can be pushed to hundreds of kHz, even
above MHz. Furthermore, with higher switching frequency, the turns number of inductor can be
reduced. This provides an opportunity to use PCB winding to build the inductors for high
frequency PFC converter. This chapter focuses on PCB winding inductor integration and EMI
noise attenuation for PFC converters. Two applications of PFC converter will be discussed. The
first one is 1kW PFC converter for server power supply, the second one is 6.6kW PFC converter
for on-board battery charger.
3.1 PCB Winding Negative Coupled Inductor with Balance Technique [41]
3.1.1 1.2kW GaN Based Totem-pole Bridgeless PFC above MHz
Comparing with the boost PFC converter, the bridgeless PFC converter has clear advantages.
It eliminates the diode rectifier bridge and reduces the conduction loss of the semiconductors.
Among the topologies of bridgeless PFC converter, the dual-boost bridgeless PFC converter (Fig.
3.2(a)) is the most popular topology in industry products. However, the totem-pole PFC converter
(Fig. 3.2(b)) is simpler and requires less semiconductor devices than dual-boost PFC converter.
With the limitation of Si MOSFET, the totem-pole PFC converter is only used in low frequency,
low power applications. When operating in continues-current mode (CCM), the turn-on loss is
very large due to the reverse recovery of the antiparallel body diode of Si MOSFET. The critical-
mode (CRM) operation can eliminate the turn-on loss but the turn-off loss of Si MOSFET is very
large. For many years, the Si MOSFET limits the use of totem-pole PFC converter [42].
Yuchen Yang Chapter 3
44
(a) Dual-boost bridgeless PFC rectifier (b) Totem-pole bridgeless PFC rectifier
Fig. 3.2 Bridgeless PFC converter topologies
With the recent development of wide-band-gap (WBG) power devices, the design of power
converters can be dramatically changed. The reverse-recovery charge of gallium-nitride (GaN)
devices is greatly reduced. Thus, the totem-pole PFC converter can operate at CCM with high
efficiency. It is demonstrated in [43], the CCM totem-pole PFC can achieve 99% efficiency.
However, in order to achieve such high efficiency, the switching frequency is limited at 50 to 100
kHz. The power density of GaN based PFC converter is similar as Si based design.
In order to get more system level benefits, the switching frequency needs to push higher. One
characteristic of GaN devices is that the turn-off loss is extremely small. As shown in Fig. 3.3, the
Eoff of GaN devices is significantly smaller than Eon. This makes the CRM a suitable operation for
high switching frequency. Fig. 3.4 shows the waveform of CRM operation. During the off-time,
after the inductor current touches zero, the inductor is resonating with device junction capacitor.
Therefore, the voltage of device VDS will resonate to zero. The device turns on after VDS reaches
zero and zero-voltage switching (ZVS) is achieved. In the CRM operation, the turn-on loss is
eliminated [44]-[47]. Although the turn-off current is very high, the small Eoff will guarantee small
switching loss.
VIN
L1
L2
D1 D2
C R VO
+
-
S1 S2DN2 DN1
Yuchen Yang Chapter 3
45
Fig. 3.3 Measured switching loss distribution of different GaN device samples
Fig. 3.4 Critical mode operation
However, the CRM totem-pole PFC converter has its own limitations. The first one is large
current ripple. In CRM PFC converter, the current ripple is two times of the average current. Such
high current ripple will lead to high DM noise and require very large DM filter. The method to
solve this issue is to use two-phase interleaved structure, as shown in Fig. 3.5. The waveforms are
0
10
20
30
40
50
60
70
80
0 5 10 15
En
erg
y (
uJ
)
Current (A)
0
0.5
1
1.5
2
2.5
3
0 5 10 15
En
erg
y (
uJ
)
Current (A)
E-GaN (sample 1)
E-GaN (sample 2)
E-GaN (sample 1)
Cascode GaN
Eon
Eoff Cascode GaN
E-GaN (sample 2)
Yuchen Yang Chapter 3
46
shown in Fig. 3.6. It can be seen that the phase current iL1 and iL2 have very large ripple. But the
ripple of input current iin is much smaller.
Fig. 3.5 Two-phase interleaved totem-pole PFC converter
Fig. 3.6 Waveforms of two-phase interleaved totem-pole PFC converter
Fig. 3.7 shows the DM noise measurement result. The blue curve is the DM noise of CRM
PFC without interleaving. The green curve is the DM noise of CRM PFC with two-phase
interleaving. It can be seen that the DM noise can be greatly reduced with two-phase interleaving.
For the one-stage DM filter with 40db/dec attenuation ability, the filter corner frequency can be
increased from 80 kHz to 600 kHz. This can greatly reduce the DM filter size.
Yuchen Yang Chapter 3
47
Fig. 3.7 DM noise measurement result
In [48], a 230 Vac to 400 Vdc two-phase interleaved 1.2 kW totem-pole PFC converter with
1 MHz switching frequency is demonstrated, as shown in Fig. 3.8. The power density is around
220W/in3. The DC-link capacitors are not included here because the DC-link capacitors design is
determined by the hold-up time requirements. The value of DC-link capacitors are related with the
design of DC-DC converters.
Fig. 3.8 Prototype of two-phase interleaved 1.2 kW totem-pole PFC converter
Yuchen Yang Chapter 3
48
Comparing with the Si MOSFET based PFC converter, the power density of GaN based
totem-pole PFC converter is 5 times higher. However, the MHz CRM PFC converter still has
several issues. One is large variation in switching frequency. The operation principle of the CRM
PFC converter is that when the inductor current touches zero, the switch turns on. And the inductor
current begin to rise. After a fixed on-time, the switch turns off, the inductor current begins to fall.
Hence, the peak current of the inductor follows the equation below.
𝑖𝑝𝑘 =𝑉𝑖𝑛𝐿𝑇𝑜𝑛 (3-1)
As long as Ton is fixed, the peak current of the inductor follows Vin. Theoretically, the
waveform of the inductor current is a series of triangle waveforms with the maximum value
described in (3-1) and minimum value as zero. Because the input voltage is AC, with fixed on-
time, the switching frequency will change with different input voltage value.
Fig. 3.9 Switching frequency variation of CRM PFC converter
Fig. 3.9 shows the switching frequency variation during a half-line cycle. It can be seen that
with a minimum switching frequency of 1 MHz, the converter can have a switching frequency
above 2MHz. This high switching frequency will cause greater loss.
Yuchen Yang Chapter 3
49
Another issue is that the CRM PFC converter cannot achieve zero voltage switching (ZVS) all
the time. When the inductor current drops to zero, the inductor will oscillate with the junction
capacitors. The peak-to-peak oscillation amplitude of the device voltage is 2(Vo-Vin). Hence, when
the input voltage is larger than 0.5Vo, the switch can only achieve valley switching instead of ZVS,
as shown in Fig. 3.10. This will bring a great deal of extra switching loss for converters with a
high switching frequency.
(a) Vin < 0.5Vo (b) Vin > 0.5Vo
Fig. 3.10 Non-ZVS at high input voltage
3.1.2 Negative Coupled Inductor in MHz CRM PFC Converter
The concept of the coupled inductor has been widely applied in multi-phase VRMs to reduce
loss and improve transient performance [49]. In [50], a coupled inductor is evaluated in an
interleaved CRM boost PFC converter. In [51], the impact of using a coupled inductor during the
resonant period is analyzed. These analyses shows that the coupled inductor is a promising
technique for high-frequency CRM PFC converters.
Fig. 3.11 shows the interleaved totem-pole PFC converter with a coupled inductor. Inductors
L1 and L2 are inversely coupled. Fig. 3.12 shows the inductor structure demonstration. The two
Yuchen Yang Chapter 3
50
inductors L1 and L2 are winded on an EI shaped magnetic core. L1 is on the left leg and L2 is on
the right leg. The red arrows indicate that the terminals of input current. The blue arrows indicate
the terminals of output current. The black dash lines are the direction of mutual flux of L1 and L2.
The green dash lines are the direction of leakage flux of L1 and L2. It can be seen that the flux
generated by L1 and L2 are on opposite direction. Hence, L1 and L2 are negatively coupled. Fig.
3.13 shows the typical waveforms for different duty cycles. In [49], the derivations of equivalent
inductances Leq1, Leq2 and Leq3 are provided. The expressions of Leq1, Leq2 and Leq3 are listed below.
Fig. 3.11 Interleaved totem-pole PFC converter with coupled inductor
Fig. 3.12 Negative coupled inductor structure demonstration
Yuchen Yang Chapter 3
51
(a) D < 0.5 (b) D > 0.5
Fig. 3.13 Inductor current waveforms of coupled inductor
𝐿𝑒𝑞1 =𝐿2−𝑀2
𝐿+𝐷
𝐷′𝑀, 𝐿𝑒𝑞2 = 𝐿 +𝑀, 𝐿𝑒𝑞3 =
𝐿2−𝑀2
𝐿+𝐷′
𝐷𝑀
(3-2)
When the input voltage is high and the duty cycle is smaller than 0.5, Leq1 determines the
phase current ripple. Hence, Leq1 can be considered as steady-state inductance when duty cycle is
smaller than 0.5. When the input voltage is low and the duty cycle is larger than 0.5, Leq3
determines the phase current ripple. Thus Leq3 is the steady-state inductance when the duty cycle
is larger than 0.5. Therefore, for a CRM PFC converter with a coupled inductor, the steady-state
inductance changes with the input voltage during a half-line cycle, as shown in Fig. 3.14. The blue
line represents the non-coupled case. The inductance is a constant value during half-line cycle.
The red, green, purple and light blue lines represent the coupled inductor with coupling coefficient
as -0.5, -0.6, -0.7 and -0.8 respectively. When duty cycle is smaller than 0.5, Lss = Leq1. When duty
cycle is larger than 0.5, Lss = Leq3. If the same steady-state inductance is kept for the non-coupled
inductor and coupled inductor at Tline/4, the coupled inductor has larger steady-state inductance
when the duty cycle approaches 0.5. Because of this, the on-time also needs to change during the
half-line cycle in order to achieve unity power factor. In the CRM operation, if the inductance is
larger, the switching frequency is lower. If the inductance is smaller, the switching frequency is
higher. Thus, the switching frequency with a coupled inductor during the half-line cycle is different
Yuchen Yang Chapter 3
52
from the non-coupled case. Fig. 3.15 shows the switching frequency during a half-line cycle with
different coupling coefficients. The blue line represents the non-coupled case. The red, green,
purple and light blue lines represent the coupled inductor with coupling coefficient as -0.5, -0.6, -
0.7 and -0.8 respectively. Because the same steady-state inductance is kept for the non-coupled
inductor and coupled inductor at Tline/4 instant, the switching frequency of non-coupled inductor
and coupled inductor are the same 1 MHz. It can be seen that the coupled-inductor CRM PFC
converter has a much lower switching frequency than the non-coupled case, except for the zero-
crossing area. Thus, the coupled inductor can reduce the average switching frequency of the CRM
PFC converter, and this can improve the converter efficiency. This is the first benefit of coupled
inductor in CRM PFC converter.
Fig. 3.14 Steady-state inductance variance during half-line cycle
Fig. 3.15 Switching frequency variance of coupled inductor during half -line cycle
Yuchen Yang Chapter 3
53
As analyzed in [51], the equivalent inductance during the resonant period is Leq4. The
expression of Leq4 is shown in (3-3). With the same steady state inductance as non-coupled inductor
(3-4), Leq4 can be expressed as (3-5). With negative coupling, it can be seen that Leq4 is always
smaller than the non-coupled inductor. Hence the resonant period with a coupled inductor is shorter
than the non-coupled case. A shorter resonant period means a larger portion of energy transferring
time. Thus, the conduction loss is smaller.
𝐿𝑒𝑞4 = 𝐿 −𝑀2
𝐿 (3-3)
𝐿𝑛𝑐 = 𝐿𝑒𝑞1 =𝐿2 −𝑀2
𝐿 +𝐷𝐷′𝑀
(3-4)
𝐿𝑒𝑞4 = 𝐿𝑛𝑐 (1 +𝐷
𝐷′𝛼) < 𝐿𝑛𝑐 (3-5)
In [51], the coupled inductor is proven to have the ability to help achieve ZVS when the duty
cycle is smaller than 0.5. As shown in Fig. 3.16 (a), the resonant amplitude without coupling is
2(Vo-Vin). Therefore, when input voltage is higher than one half of output voltage, VDS cannot
resonant to zero. With coupled inductor, the resonant amplitude is 2(Vo-Vin)(1-α), which is higher
than 2(Vo-Vin). Therefore, VDS has the ability to resonant to zero, as shown in Fig. 3.16 (b). This
is a great benefit for the CRM PFC converter. As shown in Fig. 3.17 (a), the converter cannot
achieve ZVS when input voltage is higher than 200V. The red zone is the non-ZVS range. However,
the coupled inductor, the non-ZVS range is greatly reduced, as shown in Fig. 3.17 (b). This can
help reduce the non-ZVS loss by 50%.
Yuchen Yang Chapter 3
54
(a) Non-coupled case (b) Coupled inductor case
Fig. 3.16 Coupling can help achieve ZVS
(a) Non-coupled case (b) Coupled inductor case
Fig. 3.17 Coupling can help extend ZVS range
A coupled inductor can also reduce circulating energy when the duty cycle is larger than 0.5
[51]. This characterization of coupled inductors can help improve the performance of the CRM
PFC converter. When the input voltage is low, the converter will have extra circulating energy.
The coupled inductor can reduce circulating energy during this time period. Thus the coupled
inductor can reduce conduction loss for the CRM PFC converter.
3.1.3 PCB Winding Inductor Design
In the conventional AC-DC converter, the switching frequency is around 100 kHz, the
inductor is very bulky and manually assembled. This bulky and sloppy magnetic design cannot
meet the demand of increasing efficiency and power density. With 1 MHz switching frequency,
Yuchen Yang Chapter 3
55
the inductor size can be greatly reduced, as shown in Fig. 3.18. The volume and weight of inductor
are reduced 80%. Conventional magnetic materials are not suitable for such high switching
frequency. In this high frequency inductor, P61 material is used. The core loss is shown in Fig.
3.19. It can be seen that the P61 has much lower core loss than 3F45 at 1 MHz and 2 MHz. With
the new material, the inductor can be designed smaller and more efficient.
Fig. 3.18 Size comparison between 100 kHz inductor and 1 MHz inductor
Fig. 3.19 Core loss density of 3F45 and P61
Yuchen Yang Chapter 3
56
Table 3-1 shows the loss breakdown for two non-coupled inductors. The non-coupled inductor
uses two ER23 cores. The winding is 250/46 litz wire with 10 turns for each inductor.
Table 3-1 Loss breakdown for non-coupled inductor
Winding Loss (W) Core Loss (W) Total Loss/W
2.3 2.3 4.6
Although this inductor is small and high efficient, it still need to be manually assembled and
the tolerance control is poor. In order to achieve automated manufacture and good quality control,
PCB winding magnetic is a better choice. People have been using PCB winding to build
transformers for many years [36][52]. However, PCB winding inductor for CRM PFC is still a
challenge. There is no PCB winding inductor can achieve same efficiency as conventional litz-
wire inductor.
The non-coupled inductor design above provides a baseline of inductor size and loss. The
design of a coupled inductor is shown below. Based on the previous analysis, in order to reduce
the average switching frequency, strong coupling is preferred. In this paper, α = -0.7 is chosen.
However, the typical EI core structure cannot achieve such high coupling coefficient. Therefore
the first attempt is made with a UI core (Structure I). PCB winding has been used in many
applications because it is easy to manufacture and easy to control the parasitics. Fig. 3.20 Coupled
inductor in UI core shows the structure of a coupled inductor with PCB winding in a UI core. L1
is wound on the left leg with a five-layer PCB; each layer has two turns. L2 is wound on the right
leg with 10 turns PCB winding. Table 3-2 shows the simulated loss breakdown of this coupled
inductor in the UI core structure. It can be seen that this coupled inductor has much higher winding
Yuchen Yang Chapter 3
57
loss than the non-coupled inductor. Therefore the total loss of this coupled inductor is much higher
than the non-coupled inductor. This is because for CRM operation, the AC current ripple is
dominant. High AC current ripple will lead to large AC winding loss in the PCB winding. Because
of this, PCB winding inductor is not adopted in this application.
Fig. 3.20 Coupled inductor in UI core
Table 3-2 Loss breakdown for coupled inductor Structure I
Winding Loss (W) Core Loss (W) Total Loss/W
6.7 1.3 8.0
In order to reduce the winding loss, a new structure of coupled inductor is developed
(Structure II). As shown in Fig. 3.21, the core structure is an EI core. In addition, the interleaving
concept is applied to this structure. Now there are eight turns of L1 and two turns of L2 on the left
leg, and eight turns of L2 and two turns of L1 on the right leg. Table 3-3 shows the simulated loss
breakdown of this new inductor structure. With this interleaving effect, the AC winding loss has
been greatly reduced. However, the total loss of this coupled inductor is still higher than the non-
coupled inductor. FEA simulation is conducted to study the loss, and Fig. 3.22 shows the FEA
Yuchen Yang Chapter 3
58
simulation result of the left leg. It can be seen that there is strong fringing flux near the air gap,
which will cause large loss on the winding.
Fig. 3.21 Coupled inductor with interleaved winding
Table 3-3 Loss breakdown for coupled inductor Structure II
Winding Loss (W) Core Loss (W) Total Loss/W
3.1 1.9 5.0
Fig. 3.22 Simulation of flux distribution with large fringing flux
According to [53], the winding can be cut to avoid fringing flux. Fig. 3.23 shows the new
winding structure to further reduce the winding loss (Structure III). In this structure, there is a total
of six layers of PCB winding. The bottom two layers have only one turn. The winding on the fourth
Yuchen Yang Chapter 3
59
layer has been cut to avoid the fringing flux. Table 3-4 shows the simulated loss breakdown of the
inductor in Structure III. With the new structure, the AC winding loss has been further reduced.
Fig. 3.24 Simulation result of avoiding fringing flux shows the FEA simulation result of the left
leg. It can be seen that the winding has avoid the fringing flux. The total loss of the coupled
inductor Structure III is smaller than the non-coupled inductor. The result shows that for coupled
inductor, a good PCB winding structure can achieve similar loss as litz wire. With the PCB winding,
a great amount of labor can be removed in manufacture.
Fig. 3.23 Coupled inductor structure III
Table 3-4 Loss breakdown for coupled inductor Structure III
Winding Loss (W) Core Loss (W) Total Loss/W
2.4 1.9 4.3
Fig. 3.24 Simulation result of avoiding fringing flux
Yuchen Yang Chapter 3
60
3.1.4 Balance Technique for Coupled Inductor
One drawback of this PFC converter is high CM noise. As shown in Fig. 3.25, the CM noise
of this GaN based PFC converter is much higher than conventional Si based PFC converter. This
high CM noise will impact the EMI filter design and increase the total system volume.
Fig. 3.25 CM noise comparison between GaN based PFC and Si based PFC
In [54], a balancing technique is introduced for the interleaved boost PFC converter to reduce
CM noise. The balancing technique can also be applied to the interleaved totem-pole PFC
converter with a coupled inductor. As shown in Fig. 3.26, an extra inductor is added to achieve
balance. The CM noise model is shown in Fig. 3.27. The dv/dts of the active devices are the
dominant CM noise sources. Hence they are modeled as two voltage sources: VN1 and VN2. Cb is
the parasitic capacitance between the output trace and ground, and Cd is the drain-to-ground
capacitance. According to the superposition theory, the two independent noise sources can be
studied separately. Fig. 3.28 shows the CM noise model of source VN1. Here, VN2 is a short circuit.
The balance condition for this circuit is shown in (3-6). In order to get the balance equation, the
Yuchen Yang Chapter 3
61
ratio of Z1/Z2 needs to be calculated first. From the circuit we can get the equation (3-7). Then the
ratio of Z1/Z2 can be derived as (3-8). Thus the balance equation can be derived as (3-9). It can be
derived that the balance condition for noise source VN2 is also (3-9). It is apparent that for the
coupled inductor, the mutual inductance does not impact the balance condition.
𝑍𝑎𝑍𝑏=𝐶𝑑 + 𝐶𝑏𝐶𝑑
(3-6)
{
𝑉1 =
𝑑𝑖1𝑑𝑡𝐿𝑘 +
𝑑(𝑖1 + 𝑖2)
𝑑𝑡𝑀
𝑉2 =𝑑𝑖2𝑑𝑡𝐿𝑘 +
𝑑(𝑖1 + 𝑖2)
𝑑𝑡𝑀
𝑉2 =𝑑(𝑖1 − 𝑖2)
𝑑𝑡𝐿𝑏
(3-7)
𝑍𝑎𝑍𝑏=𝑉1/𝑖1𝑉2/𝑖1
=𝐿𝑘 + 𝐿𝑏𝐿𝑏
(3-8)
𝐿𝑘𝐿𝑏=𝐶𝑏𝐶𝑑
(3-9)
Fig. 3.26 Balancing technique for interleaved totem-pole PFC converter with coupled inductor
Yuchen Yang Chapter 3
62
Fig. 3.27 CM noise model of PFC converter with balance
Fig. 3.28 CM noise model of one noise source
In [54], it is determined that in order to achieve better balance at high frequencies, the balance
inductor Lb should be coupled with the original inductor. Hence, for the coupled inductor, two
inductors L3 and L4 are applied to achieve better balance, and they are coupled with L1 and L2. Fig.
3.29 shows the circuit topology and Fig. 3.30 shows the magnetic structure. The CM noise model
is shown in Fig. 3.31. Using the superposition theory, the model of noise source VN1 is shown in
Fig. 3.32. The balance condition for this circuit is shown in (3-10). Still, the ratio of Z1/Z2 needs
to be calculated first. It is assumed that L1 and L3 are perfectly coupled, and L2 and L4 are perfectly
coupled. The number of turns of L1 and L2 is N1. The number of turns of L3 and L4 is N2. Thus,
we have the equations shown in (3-11). Then the ratio of Z1/Z2 can be derived as (3-12). Thus, the
Yuchen Yang Chapter 3
63
balance condition for source V1 is (3-13). Similarly, the balance condition for noise source VN2 is
also (3-13). The CM noise of this PFC converter can be minimized as long as this balance condition
is achieved.
𝑍1𝑍2=𝐶𝑑 + 𝐶𝑏𝐶𝑑
(3-10)
{
𝑉1𝑉3=𝑁1𝑁2
𝑉2𝑉4= −
𝑁1𝑁2
𝑉3 + 𝑉4 = 𝑉2𝑖1 = 𝑖2
(3-11)
𝑍1𝑍2=𝑉1/𝑖1𝑉2/𝑖2
=𝑁1 + 𝑁2𝑁2
(3-12)
𝑁1𝑁2
=𝐶𝑏𝐶𝑑
(3-13)
Fig. 3.29 Circuit structure of improved balance technique
Yuchen Yang Chapter 3
64
Fig. 3.30 Inductor structure demonstration of improved balance technique
Fig. 3.31 CM noise model of improved balance technique
Fig. 3.32 Effect of only noise source VN1
In the previous analysis, there are two inductors in the coupled inductor structure, and Leq1,
Leq2, Leq3 and Leq4 are derived based on that structure. The circuit performance can then be analyzed
Yuchen Yang Chapter 3
65
by using these four equivalent inductances. However, with the balance technique, there are four
inductors coupled together. In order to analyze the performance of this circuit, we need to derive
new values for Leq1, Leq2, Leq3 and Leq4. In order to derive these new values, the inductance matrix
need to be derived first. Fig. 3.33 shows the magnetic circuit. From the magnetic circuit, the
inductance matrix can be derived as (3-14). The circuit in Fig. 3.29 can be transformed in to Fig.
3.34. Then the equations for Veq1 and Veq2 can be derived as (3-15). From (3-14) and (3-15) we
can get (3-16). Then the simplified equivalent inductance Leq and Meq can be express as (3-17).
The four-inductor circuit in Fig. 3.29 can be simplified as the two-inductor circuit structure in Fig.
3.35. The self-inductance is equal to Leq, and the mutual inductance is equal to Meq. Hence, the
new expressions of Leq1, Leq2, Leq3 and Leq4 can be expressed using Leq and Meq.
[
𝑣1𝑣2𝑣3𝑣4
] = 𝐿1
[ 1 𝛼
𝑁2𝑁1
𝛼𝑁2𝑁1
𝛼 1 𝛼𝑁2𝑁1
𝑁2𝑁1
𝑁2𝑁1
𝛼𝑁2𝑁1
𝑁22
𝑁12 𝛼
𝑁22
𝑁12
𝛼𝑁2𝑁1
𝑁2𝑁1
𝛼𝑁22
𝑁12
𝑁22
𝑁12 ]
[ 𝑑𝑖1 𝑑𝑡⁄
𝑑𝑖2 𝑑𝑡⁄
𝑑𝑖3 𝑑𝑡⁄
𝑑𝑖4 𝑑𝑡⁄ ] (3-14)
{𝑣𝑒𝑞1 = 𝑣1 + 𝑣3 + 𝑣4𝑣𝑒𝑞2 = 𝑣2 + 𝑣3 + 𝑣4
(3-15)
{
𝑣𝑒𝑞1 = 𝐿1(
𝑁12 + 2(1 + 𝛼)𝑁1𝑁2 + 2(1 + 𝛼)𝑁2
2
𝑁12 ) ∙
𝑑𝑖1𝑑𝑡
+ 𝐿1(𝛼𝑁1
2 + 2(1 + 𝛼)𝑁1𝑁2 + 2(1 + 𝛼)𝑁22
𝑁12 ) ∙
𝑑𝑖2𝑑𝑡
𝑣𝑒𝑞2 = 𝐿1(𝑁12 + 2(1 + 𝛼)𝑁1𝑁2 + 2(1 + 𝛼)𝑁2
2
𝑁12 ) ∙
𝑑𝑖2𝑑𝑡
+ 𝐿1(𝛼𝑁1
2 + 2(1 + 𝛼)𝑁1𝑁2 + 2(1 + 𝛼)𝑁22
𝑁12 ) ∙
𝑑𝑖1𝑑𝑡
(3-16)
𝐿𝑒𝑞 = 𝐿1 (𝑁12 + 2(1 + 𝛼)𝑁1𝑁2 + 2(1 + 𝛼)𝑁2
2
𝑁12 )
𝑀𝑒𝑞 = 𝐿1(𝛼𝑁1
2 + 2(1 + 𝛼)𝑁1𝑁2 + 2(1 + 𝛼)𝑁22
𝑁12 )
(3-17)
Yuchen Yang Chapter 3
66
Fig. 3.33 Magnetic circuit of coupled inductor with balance
Fig. 3.34 Equivalent circuit of coupled inductor with balance
Fig. 3.35 Simplified equivalent circuit of coupled inductor with balance
In order to verify this theory, a simulation model is built where Leq and Meq of the balanced
circuit are equal to L and M of the non-balanced circuit. From Fig. 3.36, it can be seen that the
balanced circuit and non-balanced circuit have identical current waveforms.
Yuchen Yang Chapter 3
67
Fig. 3.36 Current waveforms verification
In the previous analysis, the inductor is integrated into PCB motherboard. The balance
technique can be easily implement into the PCB winding inductor. Fig. 3.37 shows the inductor
structure. By replacing the bottom layer of L1 and L2 with L3 and L4, the balance technique can be
integrated. It can be seen that L3 and L4 have only one turn. Therefore the turns ratio of N1:N2 is
9:1. Based on equation (3-11), the capacitor value needs to be changed to achieve the balance
condition. Fig. 3.38 shows that an external capacitor Cadd is applied in parallel with Cb. And the
new balance equation is shown in (3-16).
𝑁1
𝑁2=
𝐶𝑏+𝐶𝑎𝑑𝑑
𝐶𝑑 (3-18)
Fig. 3.37 Cross-section view of coupled inductor with balance technique
Yuchen Yang Chapter 3
68
Fig. 3.38 Add capacitor to maintain balance condition.
3.1.5 Hardware Demonstration
The detailed winding arrangement for the proposed PCB winding inductor is illustrated in Fig.
3.39. The windings are on the two outer legs of magnetic core. For inductor L1 on the left leg, the
current goes into the first layer, then goes to second layer through the vias. The third layer of L1
should be on the right leg. On the fifth layer, the current goes back to the left leg. Then the current
goes out from the forth layer. For inductor L2 on the right leg, the current goes into the forth layer,
then goes to fifth layer through vias. Then the current goes to the third layer on the left leg. After
two turns on the left leg, the current goes back to the second layer on the right leg and goes out
though the first layer. The bottom layer is the balance inductor L3 and L4. The terminals of L3 and
L4 are left open. This is for the easy of configuration of balance and non-balance case.
Yuchen Yang Chapter 3
69
(a) Layer 1 (b) Layer 2
(c) Layer 3 (d) Layer 4
(e) Layer 5 (f) Layer 6
Fig. 3.39 Winding arrangement for proposed PCB winding inductor
Yuchen Yang Chapter 3
70
In Fig. 3.40, the comparison between conventional discrete litz-wire inductor and proposed
PCB winding integrated inductor is demonstrated. In order to achieve similar EMI performance,
four discrete litz-wire inductors are needed. Two large inductors L1 and L2 are the main inductors
for phase 1 and phase 2. Two small inductor L3 and L4 are balance inductors for CM noise
reduction. This is a very complicated structure and implementation is labor intensive. Worse still,
the parasitic control is poor, so the EMI noise reduction is not guaranteed. On the contrary, the
proposed PCB winding inductor is able to integrate four inductors into one component. With the
winding printed on the circuit board, the inductance and parasitics are well controlled. Therefore
the circuit performance and EMI noise reduction is more consistent. Furthermore, the proposed
integrated inductor can achieve fully automated manufacturer. The elimination of labor intensive
work can further reduce the cost.
(a) Discrete litz-wire inductor (b) Integrated PCB winding inductor
Fig. 3.40 Comparison between discrete inductor and integrated inductor
Fig. 3.41 shows the hardware photo of the 1.2kW CRM PFC converter with PCB winding
inductor. The power density of this converter is 700W/in3 (without output capacitors). Fig. 3.42
shows the measured efficiency. The peak efficiency is above 99%. It is as good as the state-of-the-
L1 L2
L3 L4
Yuchen Yang Chapter 3
71
art low frequency Si based PFC converter. However, the power density of this high frequency PFC
converter is 10 times higher than the traditional low frequency design.
Fig. 3.41 Hardware photo of 1.2kW CRM PFC with PCB winding inductor
Fig. 3.42 Measured efficiency of 1.2kW PFC converter
Fig. 3.43 shows the CM noise measurement result. It can be seen that with the balance
technique, the CM noise has a 20dB reduction. This proves the effectiveness of the balance
technique. With PCB winding inductor, the parasitic can be easily controlled and therefore the
balance technique has better high frequency performance than litz wire inductor.
Yuchen Yang Chapter 3
72
With the PCB winding inductor, the CRM PFC converter can achieve high efficiency and high
power density. Furthermore, this inductor design greatly simplifies the manufacturing process by
eliminating the labor intensive procedure.
Fig. 3.43 CM noise measurement result
3.2 PCB Winding Positive Coupled Inductor with Balance Technique [55]
3.2.1 6.6kW SiC Based PFC Converter for On-board Battery Charger
With the increasing demand of energy efficiency and low combustion emissions, the Plug-in
electric vehicles (PEVs) are becoming more and more popular. There are two types of PEVs, plug-
in hybrid electric vehicles (PHEVs) and battery electric vehicles (BEVs). In the PEVs, the on-
board battery charger (OBC) is an important component. Because the charging time is of the
essence, the OBCs need to be high efficient. Furthermore, the limited space and weight in vehicle
design requires the OBCs to be light weight and compact. In commercial products, the Si based
OBCs are working at low switching frequency (<100 kHz), and have low power density (3-
Yuchen Yang Chapter 3
73
12W/in3) and low efficiency (92-94%) [56][57]. In addition, the customers show more and more
interested on bidirectional power flow of PEVs. So the PEVs can provide power to house power
grid, or power small electrical equipment in the fieldwork. However, conventional Si based OBCs
cannot achieve this function without sacrifice of efficiency and power density.
With wide-bandgap (WBG) devices, the power density and efficiency of switch mode power
supplies can be dramatically increased. With 1200V SiC MOSFET and 600V GaN devices, the
target of this 6.6 kW OBC is to increase the switching frequency to be higher than 300 kHz. The
power density should be doubled or tripled. The efficiency is higher than 95%. In addition, the
OBC should achieve bidirectional power flow. Besides the high efficiency and high power density,
the proposed design should also be able to achieve good manufacturability. Fig. 3.44 shows that
state-of-the-art industry products and the design target of proposed OBC.
Fig. 3.44 Target of high efficiency high density bidirectional OBC
WBG Device
>300kHz
>25W/in3
>95% Efficiency
Bi-directional
Efficiency
Power Density90%
94%
96%
92%
0W/in3 5W/in3 10W/in3 15W/in3 20W/in3 25W/in3
Si Device
<100kHz
3~12W/in3
92~94% Efficiency
Uni-directional
Yuchen Yang Chapter 3
74
Different from the server power supply discussed in the previous section, the OBC has a very
wide output range. Fig. 3.45 shows the lithium-ion battery charging profile. The OBC should have
the ability to provide the desired voltage and current to the battery. The battery voltage range is
250 V to 450 V. This very wide output voltage range is a challenge in the design of OBC.
Fig. 3.45 Battery charging profile
Conventionally, the OBC is consist of PFC converter and LLC DC-DC converter. With fixed
DC-link voltage, the LLC converter should be able to regulate the wide range of output voltage.
However, the optimal operating point of a LLC resonant converter is the switching frequency equal
to the resonant frequency. At this point, the LLC converter has highest efficiency. With wide
output voltage range, the switching frequency of LLC converter is dramatically deviated from the
resonant frequency to achieve the required wide range voltage gain. In this case, the efficiency of
LLC converter drops dramatically.
In order to build an on-board charger with higher efficiency and higher power density, a new
6.6kW system structure is proposed, as shown in Fig. 3.46. In this proposed structure, the first
stage is a two-phase interleaved totem-pole PFC converter with 1200V SiC devices. The second
stage is a CLLC resonant converter. The DC-link voltage is designed to have a wide range from
240
280
320
360
400
440
0
10
15
20
25
0 1h 2h 3h 4h
Ibat (A) Vbat (V)
5
Pre CC CP CV
Yuchen Yang Chapter 3
75
500V to 840V, maintaining a 2:1 ratio to the battery voltage. Thus, the resonant converter can
always working around the most efficient point. As shown in Fig. 3.47, the battery voltage is from
250 V to 450 V, the DC-link voltage is from 500 V to 840 V. Therefore, from 250 V to 420 V
battery voltage, the gain of CLLC converter is 1, which means that the CLLC converter can run at
optimum operation point. The gain range of CLLC converter is only 1 to 1.07. The CLLC converter
can achieve high efficiency all the time. However, the burden was shifted to PFC side.
Fig. 3.46 Circuit structure for 6.6kW on-board battery charger
Fig. 3.47 Gain range of proposed CLLC converter
Yuchen Yang Chapter 3
76
For this PFC converter, the totem-pole topology is still a good choice. The switching
frequency of this PFC converter is above 300 kHz. However, high switching frequency can
introduce great amount of switching loss. In order to overcome the high turn-on loss of SiC device,
critical conduction mode (CRM) is a preferred control method.
Fig. 3.48 Switching frequency variation during half line cycle
However, the CRM totem-pole PFC converter has its own limitations. The first one is large
variation of switching frequency. Fig. 3.48 shows the switching frequency variation during half
line cycle. As mentioned previously, the output voltage of PFC converter has a wide range from
500V to 840V. The full power operation range is above 600V. In this figure, the blue, red and
green solid line represent the switching frequency when output voltage equal to 600V, 700V and
800V respectively. The switching frequency range is 300 kHz to 400 kHz.
The second issue is that the CRM PFC converter has large current ripple. Although two-phase
interleaved structure is applied to reduce input current ripple, the CRM PFC converter still has
larger input current ripple compared with continues conduction mode PFC converter. Therefore,
the DM noise of CRM PFC converter is larger.
0
Fre
qu
en
cy (
Hz)
450k
350k
250k
400k
300k
Tline/4 Tline/2
Vo = 600V
Vo = 700V
Vo = 800V
Yuchen Yang Chapter 3
77
In order to solve these issues, the coupled inductor concept is applied to this PFC converter.
Furthermore, the inductor will be integrated into PCB winding to achieve high power density and
automated manufacture.
3.2.2 Negative Coupled Inductor for 6.6kW PFC
Negative coupled inductor is introduced in the Section 3.1. It is proved to achieve high power
density, high efficiency, and automated manufacture. Therefore, the concept of negative coupled
inductor is applied to this 6.6kW PFC converter. Fig. 3.49 shows the interleaved totem-pole PFC
converter with coupled inductor. Inductor L1 and L2 are negative coupled. The waveforms and
equivalent inductance of negative coupled inductor are introduced in Section 3.1.
Fig. 3.49 Negative coupled inductor for 6.6kW PFC
When input voltage is high, the duty-cycle is smaller than 0.5, Leq1 determines the phase
current ripple. Hence, Leq1 can be considered as steady state inductance when duty-cycle is smaller
than 0.5. When input voltage is low, the duty-cycle is larger than 0.5, Leq3 determines the phase
current ripple. Hence, Leq3 is the steady state inductance when duty-cycle is larger than 0.5.
Therefore, for CRM PFC converter with coupled inductor, the steady state inductance is changing
with input voltage during a half line cycle, as shown in Fig. 3.50. The dash line represents the non-
Yuchen Yang Chapter 3
78
coupled case, which is constant 10uH. The blue, red and green lines represent the steady state
inductance with negative coupled inductor for 600V, 700V and 800V output voltage respectively.
It can be seen that with negative coupling, the steady state inductance will be smaller than the non-
coupled inductance. In CRM operation, smaller inductance will lead to higher switching frequency,
larger inductance will lead to lower switching frequency.
Fig. 3.50 Steady state inductance for 6.6kW PFC with negative coupled inductor
Thus, the switching frequency variation with coupled inductor during half line cycle is
different from the non-coupled case. Fig. 3.51 shows the switching frequency during half line cycle
with different output voltage. The dash lines represent the switching frequency for non-coupled
inductor under different output voltage. The solid lines represent the switching frequency for
negative coupled inductor. It can be seen that switching frequency of negative coupling is much
higher than the non-coupled case. Therefore, the switching loss would increase a lot and negative
coupling has no benefit.
Yuchen Yang Chapter 3
79
Fig. 3.51 Switching frequency for 6.6kW PFC with negative coupled inductor
The negative coupled inductor can help reduce switching frequency in the previous 1.2kW
PFC converter. However, in this 6.6kW PFC converter, the negative coupled inductor actually
increases the switching frequency range. This is because the two applications have different input
and output voltage. The different input and output voltage will lead to different operation duty
cycle. It is known from previous analysis that the steady state inductance is impacted by duty cycle.
For the 1.2kW PFC converter, the input voltage is 230Vac, the output voltage is 400Vdc. During
half line cycle, half of the time the duty cycle is larger than 0.5, the other half of the time duty
cycle is smaller than 0.5, as shown in Fig. 3.52. For the 6.6kW PFC converter, the input voltage
is 240Vac, the output voltage is from 600Vdc to 800Vdc. When output voltage is 600V, most of
the time duty cycle is larger than 0.5. When output voltage is 700V or 800V, the duty cycle is
always larger than 0.5, as shown in Fig. 3.53. Because of this different duty cycle range, the
negative coupled inductor has opposite impact on switching frequency for the two PFC converters.
And this proves that the negative coupled inductor is not suitable in this 6.6kW PFC converter.
Yuchen Yang Chapter 3
80
Fig. 3.52 Duty cycle range for 1 kW server power supply
Fig. 3.53 Duty cycle range for 6.6 kW OBC
3.2.3 Positive Coupled Inductor for 6.6kW PFC
Then the positive coupling will be evaluated. Fig. 3.54 shows the positive coupled inductor
for PFC converter. Fig. 3.55 shows the inductor structure demonstration. The two inductors L1 and
L2 are winded on an EI shaped magnetic core. L1 is on the left leg and L2 is on the right leg. The
red arrows indicate that the terminals of input current. The blue arrows indicate the terminals of
output current. The black dash lines are the direction of mutual flux of L1 and L2. The green dash
Yuchen Yang Chapter 3
81
lines are the direction of leakage flux of L1 and L2. It can be seen that the flux generated by L1 and
L2 are same direction. Hence, L1 and L2 are positively coupled. Fig. 3.56 shows the typical inductor
waveform of positive coupling. It has three equivalent inductance. The expression of these three
equivalent inductance is shown in (3-19). L represents the self-inductance of L1 and L2. M
represents the mutual inductance of L1 and L2. When input voltage is high, the duty-cycle is smaller
than 0.5, Leq1 determines the phase current ripple. Hence, Leq1 can be considered as steady state
inductance when duty-cycle is smaller than 0.5. When input voltage is low, the duty-cycle is larger
than 0.5, Leq3 determines the phase current ripple. Hence, Leq3 is the steady state inductance when
duty-cycle is larger than 0.5.
𝐿𝑒𝑞1 =𝐿2 −𝑀2
𝐿 +𝐷𝐷′𝑀
, 𝐿𝑒𝑞2 = 𝐿 +𝑀, 𝐿𝑒𝑞3 =𝐿2 −𝑀2
𝐿 +𝐷′
𝐷 𝑀
(3-19)
Fig. 3.54 Positive coupled inductor for 6.6kW PFC
Yuchen Yang Chapter 3
82
Fig. 3.55 Positive coupled inductor structure demonstration
Fig. 3.56 Typical waveforms of positive coupled inductor
Fig. 3.57 shows the steady state inductance variation during a half line cycle for the positive
coupled inductor. The dash line represents the non-coupled inductance, which is constant 10uH.
The blue, red and green lines represent the steady state inductance with positive coupled inductor
for 600V, 700V and 800V output voltage respectively. It can be seen that with the positive coupled
inductor, the steady state inductance is higher than non-coupled inductor. Fig. 3.58 shows the
switching frequency for positive coupled inductor. The dash lines represent the switching
frequency of non-coupled inductor. The solid lines represent the switching frequency of positive
coupled inductor. With the higher steady state inductance, the positive coupled inductor has lower
switching frequency than the non-coupled inductor. Furthermore, with 700V and 800V output
Yuchen Yang Chapter 3
83
voltage, the switching frequency is almost constant. Therefore the positive coupling will reduce
the switching frequency range and thus reduce switching loss by 14%.
Fig. 3.57 Steady state inductance for 6.6kW PFC with positive coupled inductor
Fig. 3.58 Switching frequency for 6.6kW PFC with positive coupled inductor
Yuchen Yang Chapter 3
84
3.2.4 PCB Winding Inductor Design
For conventional CRM PFC converter with high current, litz wire inductor is very popular. In
CRM operation, the AC current ripple is very large. Litz wire can help reduce the skin effect and
proximity effect losses in conductors. Fig. 3.59 shows the litz wire inductors for this 6.6 kW two-
phase interleaved totem-pole PFC converter. The inductor is built with PQ 32/30 core with 3F36
material and 825/44 litz wire. The loss breakdown for the litz wire inductor is shown in Table 3-5.
Fig. 3.59 Litz wire inductors for OBC
Table 3-5 Loss breakdown for litz wire inductor
Winding Loss (W) Core Loss (W) Total Loss/W
Litz wire winding
inductor
12 13 25
However, the litz wire inductor is bulky requires labor intensive work to assemble. Therefore
the tolerance of the inductance is very large and the parasitic control is difficult. It is very difficult
to build a coupled inductor with desired inductance value with litz wire. Worse still, it is difficult
to achieve good CM noise reduction with balance technique using litz wire inductor. In order to
overcome these drawbacks, the PCB winding inductor is a preferred solution.
Yuchen Yang Chapter 3
85
However, the conventional PCB winding inductor will suffer a lot from the skin effect and
proximity effect losses in this high AC current ripple application. As shown in Fig. 3.60, two
inductors L1 and L2 are on the same EI shaped magnetic core. Therefore, the two inductors are
coupled together. Each inductor has six turns in six layer PCB board. L1 is on the left leg of the
magnetic core, L2 is on the right leg. The loss breakdown for this conventional PCB winding
inductor is shown in Table 3-6. It can be seen that the winding loss is more than 10 times higher
than litz wire inductor. This is because the PCB winding is a piece of solid copper. For this high
frequency, high AC current ripple application, the AC winding loss is very large. Therefore, this
design is not applicable in this application.
Fig. 3.60 Conventional PCB winding inductor
Table 3-6 Loss breakdown for conventional PCB winding inductor
Winding Loss (W) Core Loss (W) Total Loss/W
Conventional PCB
winding inductor
180 17 197
In order to reduce the high AC winding loss, the new PCB winding structure is proposed in
Fig. 3.61. Instead of putting all the L1 and L2 leg on two separate legs, one layer of L1 and L2
Yuchen Yang Chapter 3
86
swapped position. There is one turn of L2 on the left leg, and one turn of L1 on the right leg. With
this wingding interleaving method, the magnetic field can be reduced and the skin effect and
proximity effect losses can be greatly reduced. Table 3-7 shows the loss breakdown of this
proposed PCB winding structure. It can be seen that the winding loss has been greatly reduced.
The total loss of this PCB winding inductor now is comparable with the litz wire inductor.
Fig. 3.61 Proposed PCB winding inductor with interleave
Table 3-7 Loss breakdown for proposed PCB winding inductor with interleave
Winding Loss (W) Core Loss (W) Total Loss/W
Proposed PCB
winding inductor
24 21 45
3.2.5 Improved PCB Winding Inductor Design
Fig. 3.62 shows the prototype of the 6.6kW on-board battery charger. The upper half is PFC
converter, and the lower half is DC/DC converter. The power density of total system is 37W/in3.
It can be seen that both PFC inductor and DC/DC transformer are integrated into PCB motherboard.
Thus, all the magnetic components in this converter are built by PCB winding. Thus, the converter
can achieve fully automated manufacture.
L1
L1
L2
L2
L2 L1
Yuchen Yang Chapter 3
87
Fig. 3.62 Prototype of 6.6kW on-board battery charger
The efficiency of PFC converter is measured, as shown in Fig. 3.63. The blue curve is for litz
wire inductor version. The green dash line is the estimated converter with PCB winding inductor.
However, the measured efficiency with PCB winding inductor is shown in red line. It is much
lower than the estimation.
Fig. 3.63 Measured PFC converter efficiency
37W/in3
Yuchen Yang Chapter 3
88
In order to find out the reason for extra loss, detailed waveforms are shown in Fig. 3.64. From
the waveform it can be seen that there are spikes and ringing in the inductor current waveform.
This is because of the equivalent parallel capacitor (EPC) of the inductor. In the PCB winding
inductor, the windings have large overlapping area. Thus the EPC of PCB winding inductor is
much larger than traditional litz wire inductor. During the switching transition, the dv/dt on the
switch node will introduce displacement current through the EPC of inductor. Large EPC will lead
to large current spike. The ringing current will go through the inductor and SiC devices to generate
loss, as shown in Fig. 3.65.
Fig. 3.64 Experimental waveforms of 6.6kW PFC converter
Yuchen Yang Chapter 3
89
Fig. 3.65 Path of current spikes and ringing
In order to reduce the loss generated by EPC, the inductor design needs to be improved. Fig.
3.66 shows the inductor design parameters, winding width “a” and core length “b”. In the original
design, the ringing loss generated by EPC was not considered. In this design the EPC loss will be
taken into account.
Fig. 3.66 Inductor optimization parameters
The inductor loss breakdown for different combination of parameter a and b is shown in Fig.
3.67. The contour of total inductor loss and footprint is shown in Fig. 3.68. From this figure, the
Yuchen Yang Chapter 3
90
optimized design point can be closed. Table 3-8 shows the loss breakdown for the optimized
inductor design. It can reduce the total inductor loss by 10W.
Table 3-8 Loss breakdown for optimized inductor design
Winding Loss (W) Core Loss (W)
Ringing Loss
(W)
Total Loss/W
Optimized
Design
29 27 3 59
Original Design 24 21 36 71
Fig. 3.67 Inductor loss breakdown for inductor design
Yuchen Yang Chapter 3
91
Fig. 3.68 Contour of total loss and footprint
Fig. 3.69 Prototype Gen. 2 of 6.6kW on-board battery charger
Yuchen Yang Chapter 3
92
Fig. 3.70 Experimental waveforms of Gen. 2 6.6kW PFC converter
The Gen. 2 hardware is shown in Fig. 3.69. The experimental waveforms are shown in Fig.
3.70. It can be seen that the current spike and ringing has been greatly reduced. The efficiency is
shown in Fig. 3.71. With the new PCB winding inductor design, the efficiency of PCB converter
is improved. The peak efficiency is above 98%, which is much higher than the industry products.
Yuchen Yang Chapter 3
93
Fig. 3.71 Improved efficiency of Gen. 2 hardware
3.2.6 Balance Technique to Reduce CM Noise
Similar as previous analysis, balance technique can be applied to this PFC converter to reduce
CM noise, as shown in Fig. 3.72. Fig. 3.73 shows the CM noise model of positive coupled inductor
with balance. The two switch bridges are modeled as two noise sources. Cb is the parasitic
capacitance between output trace and ground. Cd is the drain-to-ground capacitance. In order to
achieve balance, L3 and L4 are applied to the circuit. An external capacitor Cadd is applied to the
circuit to obtain more freedom in design the balance circuit.
Yuchen Yang Chapter 3
94
Fig. 3.72 Balance technique for positive coupled inductor
Fig. 3.73 CM noise model of balanced PFC converter
Fig. 3.74 Effect of noise source VS1
Superposition theory can be applied to this circuit. As shown in Fig. 3.74, only VS1 is
considered in the circuit, VS2 is treated as short circuit. The balance condition for this circuit is
Yuchen Yang Chapter 3
95
shown in (3-20). Still, the ratio of Z1/Z2 needs to be calculated first. It is assumed that L1 and L3
are perfectly coupled, and L2 and L4 are perfectly coupled. The number of turns of L1 and L2 is N1.
The number of turns of L3 and L4 is N2. Thus, we have the equations shown in (3-21). Then the
ratio of Z1/Z2 can be derived as (3-22). Thus, the balance condition for source V1 is (3-23).
Similarly, the balance condition for noise source VN2 is also (3-23). The CM noise of this PFC
converter can be minimized as long as this balance condition is achieved.
𝑍1𝑍2=𝐶𝑑 + 𝐶𝑏 + 𝐶𝑎𝑑𝑑
𝐶𝑑
(3-20)
{
𝑉1𝑉3=𝑁1𝑁2
𝑉2𝑉4= −
𝑁1𝑁2
𝑉3 + 𝑉4 = 𝑉2𝑖1 = 𝑖2
(3-21)
𝑍1𝑍2=𝑉1/𝑖1𝑉2/𝑖2
=𝑁1 + 𝑁2𝑁2
(3-22)
𝑁1𝑁2
=𝐶𝑏 + 𝐶𝑎𝑑𝑑
𝐶𝑑
(3-23)
The positive coupled inductor can also be integrated into PCB winding. The inductor structure
is shown in Fig. 3.75. Based on the experience of previous design, the interleaved winding
structure is used to help reduce winding loss. In this inductor design, 6-layer PCB board is used
and the last layer serves as the balance inductor.
Yuchen Yang Chapter 3
96
Fig. 3.75 PCB winding positive coupled inductor with balance
Fig. 3.76 shows the CM noise reduction by using balance technique. It can be seen that there
is 23dB CM noise reduction with balance. This can greatly help the EMI filter design.
Fig. 3.76 CM noise reduction by balance technique
3.3 Conclusions
In this chapter, the proposed PCB winding inductor is demonstrated for both 1 kW server
power supply and 6.6 kW on-board battery charger. For the 1 kW server power supply, the negative
L3
L1
L2
L2
L1
L4
Yuchen Yang Chapter 3
97
coupled inductor is applied. The negative coupled inductor can help reduce the switching
frequency to reduce switching loss. In addition, the coupled inductor can help achieve ZVS to
reduce turn-on loss. In order to integrate the inductor into PCB winding, a novel inductor structure
is proposed. The inductor winding is interleaved to reduce the high AC winding loss at high
switching frequency. In order to avoid the fringing flux, the last three layers of the winding is
specially designed. The balance technique is also applied to reduce CM noise. With PCB winding,
the balance technique is easy to implement and the parasitic is easy to control. The balance
technique can help reduce the CM noise by 20 dB. The inductor is integrated in a 6-layer mother
board. The PFC converter can achieve 99% efficiency with 700 W/in3 power density. This is the
first time that the PCB winding inductor can achieve same efficiency as litz wire inductor at high
switching frequency and high AC current ripple.
For the 6.6 kW OBC, the difference between negative coupled inductor and positive coupled
inductor is analyzed. The positive coupled inductor is applied to reduce switching frequency.
Furthermore, the coupled inductor can help further reduce input current ripple to reduce DM noise
of the PFC converter. The PCB winding inductor is also implemented. Due to the large inductor
size, the EPC of the inductor is very large and introduces extra loss. By redesigning this inductor,
the EPC is greatly reduced and efficiency of the converter is improved. Balance technique is
applied to this PFC converter and can reduce 23 dB CM noise. However, it can be seen that in this
6.6 kW system, the PCB winding inductor does not have good efficiency as litz wire inductor. This
is because of large current and limited 6-layer board. In order to further improve the inductor
efficiency, more PCB layers can be applied to reduce the winding loss of PCB winding inductor.
Yuchen Yang Chapter 4
98
Chapter 4. EMI Filter Design Considerations
In today’s power electronics area, the switch mode power supplies will generate a lot of high
frequency noises. In order to prevent those noises interfere other components, the electromagnetic
interference (EMI) filter are widely used in power converters.
Fig. 4.1 State-of-the-art server power supply
Fig. 4.1 shows a typical server power supply. It has three major parts: the EMI filter, the PFC
converter and DC-DC converter. It can be seen that the EMI filter occupies about 1/4 to 1/3 total
volume and has many components. Traditionally, due to the low switching frequency and high
EMI noise, two-stage EMI filter is used to attenuate EMI noise, as shown in Fig. 4.2 It requires
three x-capacitors, four y-capacitors and two inductors. Sometimes more inductors are used for
better attenuation. It is easily noticed that the two-stage EMI filter is bulky and more expensive.
In a traditional power supply, the EMI filter occupies 1/3 of the total volume. The two-stage EMI
filter limits the power density of the power supply.
Yuchen Yang Chapter 4
99
Fig. 4.2 Two-stage EMI filter structure
In today’s power electronics products, people are always pursuing high efficiency, high power
density and low cost. With the development of GaN devices, the switching frequency of power
converters can be pushed 10 times higher. Therefore, the power density of power converters can
be increased 5 to 10 times higher than traditional Si based design [24][36][48]. As demonstrated
in Chapter 3, a high efficiency, high density 1 kW PFC converter is demonstrated. The peak
efficiency is above 99% and the power density is 350W/in3. In Chapter 2, a 1 kW DC-DC converter
is demonstrated with 97% efficiency and 700W/in3 power density. Those two converters can form
a high efficiency, high density server power supply. Therefore, the traditional two-stage EMI filter
is not suitable in this high density power converter design. A simpler, compact one-stage EMI
filter is needed.
However, extra efforts are needed in order to use the one stage EMI filter to achieve the
required attenuation. Fig. 4.3 (a) shows the measurement DM noise of this 1 kW server power
supply. It can be seen that with two-phase interleaved structure, the fundamental frequency
harmonic is canceled. The first peak is at 2 MHz. This can help increase the filter corner frequency
and reduce filter size. Fig. 4.3 (b) shows the CM noise measurement result. It can be seen that the
CM noise is very high. In order to reduce CM noise, several CM noise reduction technique is
introduced in previous chapters. In Chapter 2, the shielding technique is applied to reduce the CM
Yuchen Yang Chapter 4
100
noise of DC-DC converter. As shown in Fig. 4.3 (c), with shielding technique, the CM noise of
DC-DC converter has been greatly reduced. However, the remaining CM noise is still high. In
Chapter 3, interleaving and balance technique is used to reduce the EMI noise of PFC converter.
As shown in Fig. 4.3 (d), with balance technique, the CM noise can be greatly reduced. These EMI
noise reduction techniques makes the idea of one-stage EMI filter become possible.
(a) DM noise measurement result
(b) CM noise measurement result
Yuchen Yang Chapter 4
101
(c) CM noise with shielding technique
(d) CM noise with shielding and balance technique
Fig. 4.3 EMI noise measurement result for 1 kW server power supply
Theoretically, the EMI filter is a low-pass filter. The self parasitics and mutual coupling will
impact the filter performance. Many researches has been done to reduce the self parasitics and
mutual coupling [58]-[73]. In [67], the method of reducing mutual coupling between inductor and
capacitor is analyzed. Self parasitics of capacitor is analyzed in [68]. In [69], the mutual coupling
between two capacitors are analyzed. However, those methods is seldom used in the two-stage
EMI filter. Because those parasitics and mutual coupling will not severely impact the performance
Yuchen Yang Chapter 4
102
of two-stage EMI filter. However, in the one-stage EMI filter, the parasitics and mutual coupling
will dominating the high frequency performance.
In this chapter, the self parasitics and mutual coupling of one-stage EMI filter is analyzed.
The methods of reduction of self parasitics and mutual coupling is demonstrated. A compact and
effective single stage EMI filter design is provided.
4.1 Single Stage EMI Filter
Fig. 4.4 shows the topology of one-stage EMI filter. It has two x-capacitors, two y-capacitors
and one CM inductor. The leakage inductance of CM inductor will serve as differential mode (DM)
inductor. Comparing with the two-stage EMI filter design, the one-stage EMI filter is more
compact and simple. The designed parameter of this single stage EMI filter is: LDM = 40 uH, CX =
680 nF, LCM = 1.5 mH, CY = 3.6 nF.
Fig. 4.4 One-stage EMI filter structure
The measured EMI noise with the single-stage EMI filter is shown in Fig. 4.5. It can be seen
that the CM noise can pass the standard. At low frequency, the DM noise can pass the standard.
However, the DM noise exceeds the standard at high frequency. This means that the preliminary
design of DM filter needs to be improved.
Yuchen Yang Chapter 4
103
(a) DM noise with single stage EMI filter
(b) CM noise with single stage EMI filter
Fig. 4.5 EMI noise measurement results
In the one-stage EMI filter, the leakage inductance of CM inductor serves as the DM inductor.
The equivalent circuit for DM and CM filters are shown in Fig. 4.6 and Fig. 4.7. Two CM
capacitors are in series for DM noise, so the capacitance is only one half of the single capacitor.
Two DM inductors are in parallel for CM noise, so the equivalent inductance is only half of one
DM inductance. DM capacitors work like balance capacitors for CM noise. CM inductor is like a
short circuit for DM noise because the magnetic fluxes generated by the DM current in two coupled
windings are cancelled. Because the CM noise can pass the standard, this chapter focuses on the
DM filter.
Yuchen Yang Chapter 4
104
Fig. 4.6 Equivalent circuit for DM filter
Fig. 4.7 Equivalent circuit for CM filter
4.2 Parasitic of DM Filter
Fig. 4.8 shows the insertion voltage gain for the ideal DM filter. Ideally, the one-stage DM
filter serves as a low-pass filter. After the corner frequency, the attenuation is 60dB/dec. However,
the real DM filter will not work like the ideal case. There are self-parasitics and mutual parasitics
to impact the filter performance.
Yuchen Yang Chapter 4
105
Fig. 4.8 Insertion voltage gain for ideal DM filter
Fig. 4.9 shows the DM filter with self parasitics. For the inductor, there are equivalent parallel
capacitance (EPC) and equivalent parallel resistance (EPR). For the capacitor, there are equivalent
series inductance (ESL) and equivalent series resistance (ESR). With those self parasitics, the filter
performance on high frequency will be changed.
Fig. 4.9 DM filter with self parasitics
Yuchen Yang Chapter 4
106
Besides the self parasitics, the filter will also be impacted by mutual parasitics. Fig. 4.10 shows
the measured insertion voltage gain for the DM filter. It can be seen that the filter has much less
attenuation on the high frequency. Fig. 4.11 shows the DM filter with self and mutual parasitics.
M1 and M2 are mutual coupling between inductor and capacitor. M3 is the mutual coupling between
two capacitors. With such bad high frequency performance, the filter cannot achieve the required
attenuation. Therefore, the reduction of self parasitics and mutual couplings is critical in the one-
stage EMI filter design.
Fig. 4.10 Insertion voltage gain for actual DM filter
Fig. 4.11 DM filter with self and mutual parasitics
Yuchen Yang Chapter 4
107
4.3 Near-field Analysis for Mutual Parasitic Cancellation
Fig. 4.12 Mutual coupling between inductor and capacitor
Among all the mutual couplings discussed in the previous section, the most important
coupling is M1 and M2, which are the mutual coupling between inductor and capacitor. As shown
in Fig. 4.12. For a compact EMI filter design, the inductor and capacitor are placed close to each
other. This makes the near field coupling between inductor and capacitor very strong. In [63], the
near field measurement method is used to analyze the DM inductor. The measurement setup is
shown in Fig. 11. The Tektronix AFG3102 signal generator provides the sine wave signal with
certain frequency. The signal is then go into the inductor through the 25W power amplifier. The
near field probe is placed 5 mm above the inductor and the inductor is placed on a XY table. With
the moving of XY table, the flux on the surface will be measured in X, Y and Z directions. 10 by
10 points is measured in this surface to visualize the near field flux of the inductor. The scope is
used to monitor the flux amplitude measured by the near field probe.
Yuchen Yang Chapter 4
108
Fig. 4.13 Near-field flux measurement setup
In this work, same measurement method is used to measure the DM flux of a CM inductor.
For the DM inductor, the windings are evenly distributed on the magnetic core, as shown in Fig.
4.14 (a). However, for the CM inductor, the windings are divided into two groups, as shown in
Fig. 4.14 (b). This makes the CM inductor has different DM near field flux with the DM inductor.
(a) DM inductor structure (b) CM inductor structure
Fig. 4.14 DM inductor structure and CM inductor structure
Yuchen Yang Chapter 4
109
The measurement result is shown below. First the excitation frequency is 1 MHz, which is a
low frequency for this EMI filter inductor. Fig. 4.15 (a), (b) and (c) shows the measurement results
in X, Y and Z directions. The X and Y direction measurement result can be combines as XY plane
( 22 yx ), as shown in Fig. 4.15 (d). Because the winding is divided into two groups, the field
shows two poles. The flux is going from one pole to another pole. The Y direction shows the
maximum value at the center area. This matches out expectation. Fig. 4.16 shows the
demonstration of flux flow direction.
(a) (b)
(c) (d)
Fig. 4.15 Near-field flux measured with 1MHz excitation
Yuchen Yang Chapter 4
110
Fig. 4.16 Flux flow direction of 1 MHz excitation
Then, the excitation frequency is increased to 10 MHz. The measurement result is shown in
Fig. 4.17. There is not much difference between the 10 MHz result and 1 MHz result. However, it
is notices that the flux in one pole becomes weaker. The flux flow direction is shown in Fig. 4.18.
The flux still flows from one pole to the other pole.
Yuchen Yang Chapter 4
111
Fig. 4.17 Near field measurement result of 10 MHz
Fig. 4.18 Flux flow direction of 10 MHz excitation
Next, the excitation frequency is set to 15 MHz. The measurement result is shown in Fig. 4.19.
Comparing with the 1 MHz and 10 MHz result, at 15 MHz, the pole on top becomes very weak.
Fig. 4.20 shows the flux flow direction. The flux in more evenly distributed at top.
Yuchen Yang Chapter 4
112
Fig. 4.19 Near field measurement result of 15 MHz
Fig. 4.20 Flux flow direction of 15 MHz excitation
Yuchen Yang Chapter 4
113
Finally, the excitation frequency is increased to 25 MHz. The measurement result is shown in
Fig. 4.21. It can be seen that this result is clearly different from low frequency measurement results.
Instead having two poles, the flux now has three poles. The flux first comes from the bottom pole
and then split to go to two top poles. Fig. 4.22 shows the flux flow direction.
Fig. 4.21 Near field measurement result of 25 MHz
Yuchen Yang Chapter 4
114
Fig. 4.22 Flux flow direction of 25 MHz excitation
Based on the previous analysis, a better EMI filter performance can be achieved by better
layout design. Fig. 4.24 shows the traditional EMI filter layout. The inductor is vertically placed
on the board. The near-field flux of inductor will generate mutual flux through capacitor. In order
to reduce the mutual coupling between inductor and capacitors, the inductor is placed horizontally.
Fig. 4.23 shows different capacitor positions with horizontal inductor. There are two positions of
the capacitors, the red position and yellow position. For the red position, it can be seen that strong
Y-direction flux flows through the capacitor. Therefore, the mutual coupling will be strong. For
the yellow position, although the Y-direction flux is strong, the flux is in parallel with the capacitor.
There is no mutual coupling between inductor and capacitors. Therefore, the yellow position is
better to reduce the mutual coupling between inductor and capacitors.
Yuchen Yang Chapter 4
115
Fig. 4.23 Different capacitor position with horizontal inductor
Fig. 4.25 shows an improved EMI filter design. The inductor is laid down on the board and
the capacitors are in the red position. And Fig. 4.26 shows a second improved design. The inductor
rotates 90 degree, and the capacitors are in the yellow position.
Fig. 4.24 Traditional EMI filter layout
Yuchen Yang Chapter 4
116
Fig. 4.25 Position 1
Fig. 4.26 Position 2
Fig. 4.27 shows the DM noise measurement result with those three different filer layouts. The
traditional design has poor performance at high frequency. The improved design 1 has better high
frequency performance than traditional design. But it still cannot pass the standard at high
frequency. The improved design 2 pass the standard at both low and high frequency but with little
margin.
Yuchen Yang Chapter 4
117
Fig. 4.27 DM noise measurement result
Fig. 4.28 shows the insertion voltage gain for single stage EMI filter with M1,2 cancellation.
The improved filter has better performance than the original filter at high frequency. Hence, the
high frequency DM noise is attenuated.
Fig. 4.28 Insertion voltage gain for filter with M1,2 cancellation
Yuchen Yang Chapter 4
118
4.4 Reduction of ESL of Capacitor Branch
The ESL in the capacitor branch will reduce the high frequency attenuation ability of the filter.
In [68], a method is analyzed to reduce the ESL of capacitor branch based on circuit network
theory.
Fig. 4.29 Circuit network 1
For the circuit network shown in Fig. 4.29, the port voltage and current can be described by
the following equation with Z matrix.
(𝑉1𝑉2) = (
𝑍11 𝑍12𝑍21 𝑍22
) (𝐼1𝐼2)
(4-1)
Then the Z matrix can be calculated as:
Z = (
𝑍1 + 𝑍22
𝑍2 − 𝑍12
𝑍2 − 𝑍12
𝑍1 + 𝑍22
) (4-2)
Another circuit network can be created with same Z matrix, as shown in Fig. 4.30.
Yuchen Yang Chapter 4
119
Fig. 4.30 Circuit network 2
Because the circuit network 1 and circuit network 2 have same Z matrix, they have same
performance on two ports. Therefore, circuit network 1 is equivalent to circuit network 2. From
circuit network 1 to circuit network 2, the impedance Z1 is subtracted by Z2 and formed the
impedance (Z2-Z1)/2. Based on this circuit network transformation, the ESL of capacitor branch
can be cancelled.
Fig. 4.31 ESL cancellation of capacitor
As shown in Fig. 4.31, two capacitors are diagonally positioned, just as circuit network 1. The
two cancellation inductors are placed on the top and bottom line. According to the network theory,
if the value of cancellation inductor L is equal to the ESL of capacitor, the shunt branch in the
equivalent circuit will have no inductance.
The ESL of capacitor is usually very small at nH level. For example, for a 275 Vac 0.32 uF
capacitor, the ESL is only 9 nH. Such small inductance can be easily formed through PCB
Yuchen Yang Chapter 4
120
windings, as shown in Fig. 4.32. The two film capacitors are placed parallel. The cancellation
inductor is realized using PCB winding. The calculation of the PCB winding inductance can be
found in [74]:
𝐿 =𝜇04𝜋[4𝑙1 ln (
2𝑙1𝑤) + 4𝑙2 ln (
2𝑙2𝑤) − 4𝑙1sinh
−1 (𝑙1𝑙2) − 4𝑙2sinh
−1 (𝑙2𝑙1)
+ 8(𝑙12 + 𝑙2
2)12 − 2(𝑙1 + 𝑙2)]
(4-3)
Fig. 4.32 ESL cancellation implementation into PCB
Fig. 4.33 Calculation and simulation of PCB winding inductance
Yuchen Yang Chapter 4
121
The calculation can also be verified by FEA simulation, as shown in Fig. 4.33. Table 4-1
shows the results comparison between calculation and FEA simulation. It can be seen that with
different dimension, the inductance value matches very well.
Table 4-1 Comparison between calculation and FEA simulation
Calculation FEA simulation
Inductance 9.9 nH 9.5 nH
Fig. 4.34 shows the insertion voltage gain for the filter with ESL cancellation method. It can
be seen that with the ESL cancellation method, the high frequency performance can be improved.
However, the filter performance around 2 MHz is not improved.
Fig. 4.34 Insertion voltage gain for filter with ESL cancellation
Fig. 4.35 shows the DM noise measurement result. It can be seen that the high frequency DM
noise can be attenuated as analyzed from the insertion voltage gain. However, because the DM
filter performance is not improved around 2 MHz, the DM noise around 2 MHz is not reduced, as
shown in Fig. 4.35.
Yuchen Yang Chapter 4
122
Fig. 4.35 DM noise measurement result with ESL cancellation
4.5 Reduction of Mutual Coupling between Two Capacitors
Film capacitors are very popular for DM filters. There are two methods of producing film
capacitors: the wound technology and stacked-film technology. The wound technology is shown
in Fig. 4.36. Capacitors are made by individually rolling the metallized films or the film/foils into
cylindrical rolls and then covering them with an insulating sleeve or coating. The stacked-film
technology is shown in Fig. 4.37. Large rings of metallized film are wound onto core wheels with
diameters up to 60 cm. In this way the "master capacitors" are produced under well-defined and
constant conditions [75].
Fig. 4.36 Wound technology
Yuchen Yang Chapter 4
123
Fig. 4.37 Stacked-film technology
Fig. 4.38 Current flow path in the film capacitor
Fig. 4.38 shows the current flow path in the film capacitor. The current flows in from the wire
lead and metal contact layer on one end of the capacitor. Then the current flows through the
electrodes and dielectric films. Then the current reaches the metal contact layer and wire lead on
the other side of the capacitor. Therefore the current loop inside the capacitor is shown in Fig. 4.38.
This loop will coupled with external flux and then generate mutual coupling.
M3 represents the mutual coupling between two capacitors, as shown in Fig. 4.11. The flux
linkage is shown in Fig. 4.39. The mutual flux Φ𝑀3 couples the current loops of C1 and C2. Based
on the current direction, the two capacitors are positively coupled. [70] provides a method to
reduce the mutual coupling between two capacitors.
Yuchen Yang Chapter 4
124
Fig. 4.39 Flux linkage between two capacitors
In order to reduce the mutual coupling between the two capacitors, an additional inductor is
added in capacitor C1, as shown in Fig. 4.40 [70]. The original terminals of C1 is point A and point
B. The additional inductor has two terminals, point C and point D. Point B of C1 is connected with
point D of the additional inductor LM. Therefore, the current flows in at point C and flows out at
point A. The current direction in LM is opposite from C1. Hence, the flux generated by LM has
opposite direction as the flux generated by C1. Thus, the additional inductor and C2 are negatively
coupled. The circuit analysis is shown in Fig. 4.41. If MA equals to M3, the mutual coupling
between two capacitors can be minimized.
Fig. 4.40 Cancellation of M3 using additional inductor
Yuchen Yang Chapter 4
125
Fig. 4.41 Equivalent circuit of M3 cancellation method
Fig. 4.42 shows the insertion voltage gain of filter with M3 cancellation method. It shows that
the filter can have better attenuation from 2 MHz to 30 MHz.
Fig. 4.42 Insertion voltage gain of filter with M3 cancellation
Fig. 4.43 shows the DM noise measurement result. As predicted from the insertion voltage
gain. The DM filter is attenuated at high frequency.
Yuchen Yang Chapter 4
126
Fig. 4.43 DM noise measurement result with M3 cancellation
4.6 Single Stage EMI Filter Demonstration in 1 kW Server Power Supply
With all the proposed analysis, a 1 kW server power supply with single stage EMI filter is
demonstrated. The EMI filter uses the proposed single stage EMI filter. The volume of this single
stage EMI filter is only 20% of the traditional two-stage EMI filter. The PFC stage uses the two-
phase interleaved totem-pole structure, as analyzed in Chapter 3. With GaN devices, the switching
frequency is above 1 MHz. With the proposed PCB winding inductor structure with balance
technique, the PFC converter can achieve high efficiency, high density, low EMI noise and
automated manufacturer. The DC-DC converter is an LLC converter with matrix transformer.
Shielding method is applied to significantly reduce CM noise. All these EMI noise reduction
methods make the single stage EMI filter become possible.
Fig. 4.44 shows the DM filter performance of this single stage EMI filter. With all the mutual
coupling reduction and self-parasitic reduction methods, the single stage EMI filter can achieve
the required attenuation.
Yuchen Yang Chapter 4
127
Fig. 4.44 DM filter performance
Fig. 4.45 shows the CM filter performance of this single stage EMI filter. With the balance
and shielding technique, the CM noise has been greatly reduced. This makes the design of CM
filter very simple. It can be seen that the single stage EMI filter can achieve the required attenuation.
Fig. 4.45 CM filter performance
Yuchen Yang Chapter 4
128
4.7 Conclusions
Traditionally, two-stage EMI filter is applied in the 1 kW server power supply. The two-stage
EMI filter is bulky and expensive. With the GaN devices, the switching frequency of server power
supply is increased to above 1 MHz. This can greatly reduce the converter volume and improve
the power density. The bulky two-stage EMI filter is not suitable for this high density converter
design. In order to use simple one-stage EMI filter, several EMI noise reduction method is applied.
For the PFC converter, the balance technique is applied to reduce CM noise. Interleaving is applied
to reduce DM noise. For the isolated DC-DC converter, shielding technique is applied to reduce
CM noise through the transformer. With these techniques, the EMI noise of the converter has been
greatly reduced. Therefore, one-stage EMI filter can be applied to achieve desired attenuation.
However, the traditional one-stage EMI filter cannot meet the EMI requirements. The self
parasitics and mutual coupling will impact the high frequency performance of EMI filter. This
impact is more severe in one-stage EMI filter than in two-stage EMI filter. In this chapter, the self
parasitics and mutual coupling of EMI filter components are analyzed. Near field measurement is
applied to visualize the flux near filter components. A better filter arrangement is proposed to
reduce the mutual coupling between inductor and capacitors. The method to reduce the self
parasitics and mutual coupling between capacitors are then analyzed. A simple and compact one-
stage EMI filter is demonstrated on a 1 kW server power supply.
Yuchen Yang Chapter 5
129
Chapter 5. Conclusions and Future Work
For today’s power electronics products, with given quality and reliability, high efficiency,
high power density and low cost are getting more and more attention. With recent advances made
in wide-bandgap (WBG) devices, the design of power converters can be dramatically changed.
With much faster switching speed and lower switching loss of WBG devices, the switching
frequency of power converters can be increased 10 times higher than Si MOSFET design. The
power density of converters can be dramatically increased. This brings challenges to the EMI
design. With fast switching speed, the dv/dt is much higher, which will lead to high common mode
noise. The PCB winding transformer is high efficiency and high density. But the inter-winding
capacitance is very large. Therefore, the EMI noise will be larger. Conventional EMI filter design
uses two-stage structure. It is very bulky and expensive. With high density power converters, the
two-stage EMI filter will occupies a lot volume and greatly reduce the system power density.
To solve these issues, this dissertation proposes several solutions. In order to reduce the CM
noise of isolated converters, shielding method is proposed. The PCB winding transformer suffers
from large inter-winding capacitance. A novel shielding method is proposed to help reduce the
CM noise so that it can be used in off-line power supplies. The design methodology of shielding
is analyzed. With the proposed shielding method, the CM noise can be reduced more than 20 dB.
More importantly, the shielding remains effective at very high frequency up to 30 MHz. The
proposed shielding method can be implemented to different converter topologies.
In the PFC converter, in order to achieve high efficiency, CRM is applied. However, with
CRM operation, the switching frequency has a wide range and the inductor current ripple is large.
This dissertation proposes a novel PCB integrated coupled inductor structure to solve these issues.
Yuchen Yang Chapter 5
130
Negative coupled inductor is applied to 1 kW server PFC converter. It can greatly reduce the
switching frequency to reduce switching related loss. It can help achieve ZVS to reduce turn-on
loss. A novel inductor structure is proposed that for the first time, the CRM PFC inductor can be
integrated into PCB with similar efficiency as litz wire inductor. The PCB winding inductor has
much higher power density and good manufacturability. Furthermore, balance technique can be
implemented into PCB winding inductor to reduce CM noise of PFC converter. With PCB winding,
the parasitic is easy to control and balance technique has better performance. It can achieve more
than 20 dB attenuation. Then the positive coupled inductor is applied to 6.6 kW OBC PFC. The
difference of negative coupling and positive coupling is analyzed. The positive coupled inductor
can help reduce the switching frequency for 6.6 kW OBC PFC converter. The inductor is also
integrated into PCB motherboard. However, due to large inductor size, the EPC of the inductor is
large. This introduces extra loss. A new design methodology is proposed to optimize the inductor.
With the improved inductor design, the PFC converter has better efficiency. Balance technique is
also applied to help reduce CM noise.
For 1 kW server power supply, two-stage EMI filter is a popular solution. However, it is bulky,
expensive and not suitable for today’s high density power converters. This dissertation proposes
the design of one-stage EMI filter. Thanks to the previous EMI noise reduction method, one-stage
EMI filter can have the opportunity to achieve the required EMI attenuation. However, the
parasitics and mutual coupling will severely impact the one-stage EMI filter. This dissertation
analyzes the self parasitics and mutual coupling. The near field measurement method is applied to
visualize the flux near filter components. The method to reduce the self parasitics and mutual
coupling between filter components are demonstrated. Finally, a simple, effective one-stage EMI
filter is demonstrated in the 1 kW server power supply system.
Yuchen Yang Chapter 5
131
With the switching frequency of power converters above MHz, new magnetic materials are
desperately need. Most today’s magnetic materials are suitable for low frequency. They will suffer
from high core loss at high frequency. The materials limit the magnetic design to be more efficient
and more compact.
The PCB winding inductor has been applied to 1 kW server PFC and 6.6 kW OBC PFC
converter. The PCB winding inductor performs well in low power applications. But in higher
power, the PCB winding inductor faces some issues. Further efforts need to be spend on how to
design PCB winding inductors at high current condition. What is the limitation of PCB winding
inductor should be investigated.
The balance technique can achieve good CM noise reduction at low frequency. But at high
frequency, the balance technique is not so effective. It is meaningful to investigate how to improve
the high frequency performance of balance technique.
This dissertation only studies the conducted EMI noise. The radiated EMI noise is also an
important topic. As the switching frequency goes beyond MHz, the radiated EMI noise would
become severe. It is worth to take effort to study the radiated EMI noise for high frequency
converters.
132
Reference
[1] International Electrotechnical Vocabulary IEV 161-01-07.
[2] International Electrotechnical Vocabulary IEV 161-01-06.
[3] Agilent Technologies: “AN150 Spectrum Analysis Basics”.
[4] EN55022: Information Technology Equipment - Radio Disturbance Characteristics – Limits
and Methods of Measurements.
[5] Wang S. Characterization and Cancellation of High-Frequency Parasitics for EMI Filters
and Noise Separators in Power Electronics Applications [Ph.D. Dissertation]. Blacksburg:
Department of Electrical and Computer Engineering, Virginia Tech, 2005.
[6] Shuo Wang; Lee, F.C., "Common-Mode Noise Reduction for Power Factor Correction
Circuit With Parasitic Capacitance Cancellation," Electromagnetic Compatibility, IEEE
Transactions on , vol.49, no.3, pp.537,542, Aug. 2007.
[7] Shuo Wang; Pengju Kong; Lee, F.C., "Common Mode Noise Reduction for Boost
Converters Using General Balance Technique," Power Electronics, IEEE Transactions on ,
vol.22, no.4, pp.1410,1416, July 2007.
[8] Pengju Kong (2009). Common mode EMI noise reduction techniques in switch mode power
supplies (Doctoral Dissertation).
[9] Pengju Kong; Shuo Wang; Lee, F.C., "Improving balance technique for high frequency
common mode noise reduction in boost PFC converters," Power Electronics Specialists
Conference, 2008. PESC 2008. IEEE , vol., no., pp.2941,2947, 15-19 June 2008.
[10] Pengju Kong; Shuo Wang; Lee, F.C., "Common Mode EMI Noise Suppression for
Bridgeless PFC Converters," Power Electronics, IEEE Transactions on , vol.23, no.1,
pp.291,297, Jan. 2008.
[11] Pengju Kong; Shuo Wang; Lee, F.C.; Chuanyun Wang, "Common-Mode EMI Study and
Reduction Technique for the Interleaved Multichannel PFC Converter," Power Electronics,
IEEE Transactions on , vol.23, no.5, pp.2576,2584, Sept. 2008.
133
[12] Rockhill A, Lipo T A, Julian A L. High Voltage Buck Converter Topology for Common
Mode Voltage Reduction. Proceedings of IEEE APEC 1998:940-943.
[13] Crebier J, Ferrieux J. PFC Full Bridge Rectifiers EMI Modeling and Analysis-common
Mode Disturbance Reduction. IEEE Transactions on Power Electronics 2004,19(2):378-
387.
[14] Wu H, Chen X, Zhou M, Zeng J, Ying J. A Novel Control Method of Interleaved Two-
Transistor Forward Converter. Proceedings of IEEE PESC 2007:2812-2818.
[15] Wu X, Poon F, Lee C M, Pong M H, Qian Z. A Study of Common Mode Noise in Switching
Power Supply from a Current Balancing Viewpoint. Proceedings of IEEE PEDS
1999(2):621-625.
[16] Hofsajer I W, Fricker K, Crosato M O. Reduction of EMI in Planer Transformers by Charge
Balancing. IEEE Electronics Letters 2002,38(25):1712-1714.
[17] Cochrane D, Chen D, Boroyevic D. Passive Cancellation of Common-mode Noise in Power
Electronic Circuits. IEEE Transactions on Power Electronics 2003,18(3):756-763.
[18] Crosato M O, Hofsajer I W. Minimising Conducted Common Mode EMI by Charge
Balancing in a Nonisolated DC-DC Converter. Proceedings of IEEE PESC 2004(4):3146-
3151.
[19] Shoyama M, Li G, Ninomiya T. Balanced Switching Converter to Reduce Common-Mode
Conducted Noise. IEEE Transactions on Industrial Electronics 2003,50(6):1095-1099.
[20] Chan Woong Park, "Method and apparatus for substantially reducing electrical earth
displacement current flow generated by wound components without requiring additional
windings," U.S. Patent 7109836 B2, 2006.
[21] S. Lin, M. Zhou, W. Chen, and J. Ying, “Novel methods to reduce common-mode noise
based on noise balance,” in Proc. 2006 IEEE Power Electron.Spec. Conf., pp. 2728–2733.
[22] Lee, F.C.; Barbosa, P.; Peng Xu; Jindong Zhang; Yang, B.; Canales, Francisco, "Topologies
and design considerations for distributed power system applications," Proceedings of the
IEEE , vol.89, no.6, pp.939,950, Jun 2001.
134
[23] M. M. Jovanovic "Technology drivers and trends in power supplies for
computer/telecom", Proc. APEC, 2006.
[24] X. Huang, J. Feng, W. Du, F. C. Lee and Q. Li, "Design consideration of MHz active clamp
flyback converter with GaN devices for low power adapter application," 2016 IEEE Applied
Power Electronics Conference and Exposition (APEC), Long Beach, CA, 2016, pp. 2334-
2341. G. Huang, A. J. Zhang, and Y. Gu, “LLC series resonant DC-to-DC converter,” U.S.
Patent 6 344 979, Feb. 5, 2002.
[25] Y. Yang, D. Huang, F. C. Lee and Q. Li, "Transformer shielding technique for common
mode noise reduction in isolated converters," 2013 IEEE Energy Conversion Congress and
Exposition, Denver, CO, 2013, pp. 4149-4153.
[26] B. Yang , Y. Ren and F. C. Lee "Integrated magnetic for LLC resonant converter", Proc.
IEEE APEC, pp.346 -351 2002.
[27] B. Yang , F. C. Lee , A. J. Zhang and G. Huang "LLC resonant converter for front end dc/dc
conversion", Proc. IEEE APEC, pp.1108 -1112 2002.
[28] Bo Yang, “Topology investigation for front end DC/DC power conversion for distributed
power system,” Ph.D. dissertation, Dept. ECE, Virginia Tech, Blacksburg, VA, USA, 2003.
[29] Y. Gu , Z. Lu , L. Hang , Z. Qian and G. Huang "Three-level LLC series resonant DC/DC
converter", IEEE Trans. Power Electron., vol. 20, no. 4, pp.781 -789 2005.
[30] D. Fu , Y. Liu , F. C. Lee and M. Xu "A novel driving scheme for synchronous rectifiers in
LLC resonant converters", IEEE Trans. Power Electron., vol. 24, no. 5, pp.1321 -1329
2009.
[31] Xinke Wu, Guichao Hua, Junming Zhang, and Zhaoming Qian, “A new current-driven
synchronous rectifier for series–parallel resonant (LLC) DC–DC converter,” IEEE Trans.
Ind. Electron., vol. 58, no. 1, pp. 289–297, Jan. 2011.
[32] Xinke Wu, Chen Hu, Junming Zhang, and Chen Zhao, “Series–parallel autoregulated
charge-balancing rectifier for multioutput light-emitting diode driver,” IEEE Trans. Ind.
Electron., vol. 61, no. 3, pp. 1262–1268, Mar.2014.
135
[33] Edward Herbert, "Design and Application of Matrix Transformers and Symmetrical
Converters," a tutorial presented at the High Frequency Power Conversion Conference '90,
Santa Clara, CA, May 11, 1990.
[34] K. Kit Sum, D. Trevor Holmes, "Novel Low Profile Matrix Transformers for High Density
Power Conversion," Power Conversion and Intelligent Motion, September, 1988.
[35] D. Reusch and F. C Lee "High frequency bus converter with low loss integrated matrix
transformer", Proc. IEEE APEC 2012, pp.1392 -1397.
[36] Daocheng Huang; Shu Ji; Lee, F.C., "LLC Resonant Converter With Matrix Transformer,"
Power Electronics, IEEE Transactions on , vol.29, no.8, pp.4339,4347, Aug. 2014.
[37] International Rectifier. (Feb. 2010). GaNpowIR—An introduction [On-line]. Available:
http://www.IRF.com.
[38] Efficient Power Conversion. EPC1015—Enhancement mode power transistor [Online].
Available: www.EPC.com.
[39] N. Dai and F. C. Lee "Edge effect analysis in a high-frequency transformer", Proc. 25th
Annu. IEEE Power Electron. Spec. Conf., Rec., pp.850 -855 1994.Laszol Huber, Yungtaek
Jang, Milan M. Jovanonic, “Performance Evaluation of Bridgeless PFC Boost Rectifiers,”
IEEE Transaction on Power Electronics, May 2008.
[40] Y. Yang, D. Huang, F. C. Lee and Q. Li, "Analysis and reduction of common mode EMI
noise for resonant converters," 2014 IEEE Applied Power Electronics Conference and
Exposition - APEC 2014, Fort Worth, TX, 2014, pp. 566-571.
[41] Y. Yang, M. Mu, Z. Liu, F. C. Lee and Q. Li, "Common mode EMI reduction technique for
interleaved MHz critical mode PFC converter with coupled inductor," 2015 IEEE Energy
Conversion Congress and Exposition (ECCE), Montreal, QC, 2015, pp. 233-239.
[42] L. Huber, Y. Jang and M. M. Jovanovic, "Performance Evaluation of Bridgeless PFC Boost
Rectifiers," in IEEE Transactions on Power Electronics, vol. 23, no. 3, pp. 1381-1390, May
2008.
[43] L. Zhou, Y. Wu, J. Honea and Z. Wang, "High-efficiency True Bridgeless Totem Pole PFC
based on GaN HEMT: Design Challenges and Cost-effective Solution," Proceedings of
136
PCIM Europe 2015; International Exhibition and Conference for Power Electronics,
Intelligent Motion, Renewable Energy and Energy Management, Nuremberg, Germany,
2015, pp. 1-8.
[44] X. Huang, Z. Liu, Q. Li, and F C. Lee, “Evaluation and application of 600 V GaN HEMT in
cascode structure,” IEEE Trans. Power Electron., vol. 29, no. 5, pp. 2453–2461, May 2014.
[45] X. Huang, Q. Li, Z. Liu, and F C. Lee, “Analytical loss model of high voltage GaN HEMT
in cascode configuration,” IEEE Trans. Power Electron., vol. 29, no. 5, pp. 2208–2219, May
2014.
[46] Z. Liu, X. Huang, F. C. Lee, and Q. Li, “Package parasitic inductance extraction and
simulation model development for the high-voltage cascode GaN HEMT,” IEEE Trans.
Power Electron., vol. 29, no. 4, pp. 1977–1985, April 2014
[47] X. Huang, W. Du, F. C. Lee, Q. Li, Z. Liu, "Avoiding Si MOSFET avalanche and achieving
zero-voltage switching for cascode GaN devices," IEEE Trans. Power Electron., vol. 31, no.
1, pp. 593–600, Jan. 2016.
[48] Z. Liu, F. C. Lee, Q. Li and Y. Yang, "Design of GaN-Based MHz Totem-Pole PFC
Rectifier," in IEEE Journal of Emerging and Selected Topics in Power Electronics, vol. 4,
no. 3, pp. 799-807, Sept. 2016.
[49] P. Wong, P. Xu, P. Yang, and F. C. Lee, “Performance improvements of interleaving VRMs
with coupling inductors,” IEEE Trans. Power Electron., vol. 16, no. 4, pp. 499–507, Jul.
2001.
[50] Fei Yang; Xinbo Ruan; Yang Yang; Zhihong Ye, "Interleaved Critical Current Mode Boost
PFC Converter With Coupled Inductor," Power Electronics, IEEE Transactions on, vol.26,
no. 9, pp. 2404, 2413, Sept. 2011.
[51] Xiucheng Huang; Lee, F.C.; Qiang Li; Weijing Du, "MHz GaN-based interleaved CRM bi-
directional buck/boost converter with coupled inductor," Applied Power Electronics
Conference and Exposition (APEC), 2015 IEEE, pp. 2075, 2082, 15-19 March 2015.
[52] M. Mu and F. C. Lee. “Design and optimization of a 380V-12V high-frequency, high-
Current LLC converter with GaN devices and planar matrix transformers,” IEEE Journal of
Emerging and Selected Topics in Power Electronics, vol. 4, no. 3, pp. 854-862, Sep. 2016.
137
[53] Dianbo Fu; Lee, F.C.; Shuo Wang, "Investigation on transformer design of high frequency
high efficiency dc-dc converters," Applied Power Electronics Conference and Exposition
(APEC), 2010 Twenty-Fifth Annual IEEE, pp. 940, 947, 21-25 Feb. 2010.
[54] Pengju Kong; Shuo Wang; Lee, F.C.; Chuanyun Wang, "Common-Mode EMI Study and
Reduction Technique for the Interleaved Multichannel PFC Converter," Power Electronics,
IEEE Transactions on, vol. 23, no. 5, pp. 2576, 2584, Sept. 2008.
[55] Y. Yang, Z. Liu, F. C. Lee and Q. Li, "Multi-phase coupled and integrated inductors for
critical conduction mode totem-pole PFC converter," 2017 IEEE Applied Power Electronics
Conference and Exposition (APEC), Tampa, FL, 2017, pp. 1804-1809.
[56] J. S. Lai, L. Zhang, Z. Zahid, N. H. Tseng, C. S. Lee and C. H. Lin, "A high-efficiency 3.3-
kW bidirectional on-board charger," 2015 IEEE 2nd International Future Energy
Electronics Conference (IFEEC), Taipei, 2015, pp. 1-5.
[57] J. S. Lai, L. Zhang, Z. Zahid, N. H. Tseng, C. S. Lee and C. H. Lin, "A high-efficiency 3.3-
kW bidirectional on-board charger," 2015 IEEE 2nd International Future Energy
Electronics Conference (IFEEC), Taipei, 2015, pp. 1-5.
[58] W. Shuo, et al., "Parasitic Effects of Grounding Paths on Common-Mode EMI Filter's
Performance in Power Electronics Systems," Industrial Electronics, IEEE Transactions on,
vol. 57, pp. 3050-3059, 2010.
[59] W. Shuo, et al., "Design of Inductor Winding Capacitance Cancellation for EMI
Suppression," Power Electronics, IEEE Transactions on, vol. 21, pp. 1825-1832, 2006.
[60] W. Shuo, et al., "A Study of Integration of Parasitic Cancellation Techniques for EMI Filter
Design With Discrete Components," Power Electronics, IEEE Transactions on, vol. 23, pp.
3094-3102, 2008.
[61] C. Rengang, et al., "Improving the Characteristics of integrated EMI filters by embedded
conductive Layers," Power Electronics, IEEE Transactions on, vol. 20, pp. 611-619, 2005.
[62] J. D. van Wyk, et al., "Integrating active, passive and EMI-filter functions in power
electronics systems:a case study of some technologies," Power Electronics, IEEE
Transactions on, vol. 20, pp. 523-536, 2005.
138
[63] W. Ruxi, et al., "High power density EMI filter design with consideration of self-parasitic,"
in Applied Power Electronics Conference and Exposition (APEC), 2012 Twenty-Seventh
Annual IEEE, 2012, pp. 2285-2289.
[64] T. De Oliveira, et al., "PEEC-models for EMC filter layout optimization," in Integrated
Power Electronics Systems (CIPS), 2010 6th International Conference on, 2010, pp. 1-6.
[65] T. De Oliveira, et al., "Automatic layout optimization of an EMC filter," in Energy
Conversion Congress and Exposition (ECCE), 2010 IEEE, 2010, pp. 2679-2685.
[66] S. P. Weber, et al., "Predicting parasitics and inductive coupling in EMI-filters," in Applied
Power Electronics Conference and Exposition, 2006. APEC '06. Twenty-First Annual IEEE,
2006, p. 4 pp.
[67] Shuo Wang, Rengang Chen, J. D. Van Wyk, F. C. Lee and W. G. Odendaal, "Developing
parasitic cancellation technologies to improve EMI filter performance for switching mode
power supplies," in IEEE Transactions on Electromagnetic Compatibility, vol. 47, no. 4, pp.
921-929, Nov. 2005.
[68] S. Wang, F. C. Lee and W. G. Odendaal, "Cancellation of capacitor parasitic parameters for
noise reduction application," in IEEE Transactions on Power Electronics, vol. 21, no. 4, pp.
1125-1132, July 2006.
[69] W. Shuo, et al., "Improvement of EMI filter performance with parasitic coupling
cancellation," Power Electronics, IEEE Transactions on, vol. 20, pp. 1221-1228, 2005.
[70] W. Shuo, et al., "Effects of parasitic parameters on EMI filter performance," Power
Electronics, IEEE Transactions on, vol. 19, pp. 869-877, 2004.
[71] M. Hartmann, et al., "EMI Filter Design for a 1 MHz, 10 kW Three-Phase/Level PWM
Rectifier," Power Electronics, IEEE Transactions on, vol. 26, pp. 1192-1204, 2011.
[72] J. Aime, et al., "Determination of the layout influence on the effectiveness of a three-phase
common mode filter by using equivalent circuits and PSpice," in Industrial Electronics,
2008. ISIE 2008. IEEE International Symposium on, 2008, pp. 1-6.
[73] C. Wei, et al., "Near Field Coupling Effects on Conducted EMI in Power Converter," in
Power Electronics Specialists Conference, 2006. PESC '06. 37th IEEE, 2006, pp. 1-6.